Computer Vision and Pattern Recognition (cs.CV)

Abstract: Vision-language models (VLMs), such as CLIP and ALIGN, are generally trained on datasets consisting of image-caption pairs obtained from the web. However, real-world multimodal datasets, such as healthcare data, are significantly more complex: each image (e.g. X-ray) is often paired with text (e.g. physician report) that describes many distinct attributes occurring in fine-grained regions of the image. We refer to these samples as exhibiting high pairwise complexity, since each image-text pair can be decomposed into a large number of region-attribute pairings. The extent to which VLMs can capture fine-grained relationships between image regions and textual attributes when trained on such data has not been previously evaluated. The first key contribution of this work is to demonstrate through systematic evaluations that as the pairwise complexity of the training dataset increases, standard VLMs struggle to learn region-attribute relationships, exhibiting performance degradations of up to 37% on retrieval tasks. In order to address this issue, we introduce ViLLA as our second key contribution. ViLLA, which is trained to capture fine-grained region-attribute relationships from complex datasets, involves two components: (a) a lightweight, self-supervised mapping model to decompose image-text samples into region-attribute pairs, and (b) a contrastive VLM to learn representations from generated region-attribute pairs. We demonstrate with experiments across four domains (synthetic, product, medical, and natural images) that ViLLA outperforms comparable VLMs on fine-grained reasoning tasks, such as zero-shot object detection (up to 3.6 AP50 points on COCO and 0.6 mAP points on LVIS) and retrieval (up to 14.2 R-Precision points).

Abstract: Hand-object interaction understanding and the barely addressed novel view synthesis are highly desired in the immersive communication, whereas it is challenging due to the high deformation of hand and heavy occlusions between hand and object. In this paper, we propose a neural rendering and pose estimation system for hand-object interaction from sparse views, which can also enable 3D hand-object interaction editing. We share the inspiration from recent scene understanding work that shows a scene specific model built beforehand can significantly improve and unblock vision tasks especially when inputs are sparse, and extend it to the dynamic hand-object interaction scenario and propose to solve the problem in two stages. We first learn the shape and appearance prior knowledge of hands and objects separately with the neural representation at the offline stage. During the online stage, we design a rendering-based joint model fitting framework to understand the dynamic hand-object interaction with the pre-built hand and object models as well as interaction priors, which thereby overcomes penetration and separation issues between hand and object and also enables novel view synthesis. In order to get stable contact during the hand-object interaction process in a sequence, we propose a stable contact loss to make the contact region to be consistent. Experiments demonstrate that our method outperforms the state-of-the-art methods. Code and dataset are available in project webpage https://iscas3dv.github.io/HO-NeRF.

Abstract: Transformers have demonstrated tremendous success not only in the natural language processing (NLP) domain but also the field of computer vision, igniting various creative approaches and applications. Yet, the superior performance and modeling flexibility of transformers came with a severe increase in computation costs, and hence several works have proposed methods to reduce this burden. Inspired by a cost-cutting method originally proposed for language models, Data Multiplexing (DataMUX), we propose a novel approach for efficient visual recognition that employs additional dim1 batching (i.e., concatenation) that greatly improves the throughput with little compromise in the accuracy. We first introduce a naive adaptation of DataMux for vision models, Image Multiplexer, and devise novel components to overcome its weaknesses, rendering our final model, ConcatPlexer, at the sweet spot between inference speed and accuracy. The ConcatPlexer was trained on ImageNet1K and CIFAR100 dataset and it achieved 23.5% less GFLOPs than ViT-B/16 with 69.5% and 83.4% validation accuracy, respectively.

Abstract: Few-shot segmentation (FSS) is a dense prediction task that aims to infer the pixel-wise labels of unseen classes using only a limited number of annotated images. The key challenge in FSS is to classify the labels of query pixels using class prototypes learned from the few labeled support exemplars. Prior approaches to FSS have typically focused on learning class-wise descriptors independently from support images, thereby ignoring the rich contextual information and mutual dependencies among support-query features. To address this limitation, we propose a joint learning method termed Masked Cross-Image Encoding (MCE), which is designed to capture common visual properties that describe object details and to learn bidirectional inter-image dependencies that enhance feature interaction. MCE is more than a visual representation enrichment module; it also considers cross-image mutual dependencies and implicit guidance. Experiments on FSS benchmarks PASCAL-$5^i$ and COCO-$20^i$ demonstrate the advanced meta-learning ability of the proposed method.

Abstract: Cross-modal garment synthesis and manipulation will significantly benefit the way fashion designers generate garments and modify their designs via flexible linguistic interfaces.Current approaches follow the general text-to-image paradigm and mine cross-modal relations via simple cross-attention modules, neglecting the structural correspondence between visual and textual representations in the fashion design domain. In this work, we instead introduce DiffCloth, a diffusion-based pipeline for cross-modal garment synthesis and manipulation, which empowers diffusion models with flexible compositionality in the fashion domain by structurally aligning the cross-modal semantics. Specifically, we formulate the part-level cross-modal alignment as a bipartite matching problem between the linguistic Attribute-Phrases (AP) and the visual garment parts which are obtained via constituency parsing and semantic segmentation, respectively. To mitigate the issue of attribute confusion, we further propose a semantic-bundled cross-attention to preserve the spatial structure similarities between the attention maps of attribute adjectives and part nouns in each AP. Moreover, DiffCloth allows for manipulation of the generated results by simply replacing APs in the text prompts. The manipulation-irrelevant regions are recognized by blended masks obtained from the bundled attention maps of the APs and kept unchanged. Extensive experiments on the CM-Fashion benchmark demonstrate that DiffCloth both yields state-of-the-art garment synthesis results by leveraging the inherent structural information and supports flexible manipulation with region consistency.

Abstract: Modern computer vision services often require users to share raw feature descriptors with an untrusted server. This presents an inherent privacy risk, as raw descriptors may be used to recover the source images from which they were extracted. To address this issue, researchers recently proposed privatizing image features by embedding them within an affine subspace containing the original feature as well as adversarial feature samples. In this paper, we propose two novel inversion attacks to show that it is possible to (approximately) recover the original image features from these embeddings, allowing us to recover privacy-critical image content. In light of such successes and the lack of theoretical privacy guarantees afforded by existing visual privacy methods, we further propose the first method to privatize image features via local differential privacy, which, unlike prior approaches, provides a guaranteed bound for privacy leakage regardless of the strength of the attacks. In addition, our method yields strong performance in visual localization as a downstream task while enjoying the privacy guarantee.

Abstract: Visual affordance segmentation identifies the surfaces of an object an agent can interact with. Common challenges for the identification of affordances are the variety of the geometry and physical properties of these surfaces as well as occlusions. In this paper, we focus on occlusions of an object that is hand-held by a person manipulating it. To address this challenge, we propose an affordance segmentation model that uses auxiliary branches to process the object and hand regions separately. The proposed model learns affordance features under hand-occlusion by weighting the feature map through hand and object segmentation. To train the model, we annotated the visual affordances of an existing dataset with mixed-reality images of hand-held containers in third-person (exocentric) images. Experiments on both real and mixed-reality images show that our model achieves better affordance segmentation and generalisation than existing models.

Abstract: Learning object segmentation in image and video datasets without human supervision is a challenging problem. Humans easily identify moving salient objects in videos using the gestalt principle of common fate, which suggests that what moves together belongs together. Building upon this idea, we propose a self-supervised object discovery approach that leverages motion and appearance information to produce high-quality object segmentation masks. Specifically, we redesign the traditional graph cut on images to include motion information in a linear combination with appearance information to produce edge weights. Remarkably, this step produces object segmentation masks comparable to the current state-of-the-art on multiple benchmarks. To further improve performance, we bootstrap a segmentation network trained on these preliminary masks as pseudo-ground truths to learn from its own outputs via self-training. We demonstrate the effectiveness of our approach, named LOCATE, on multiple standard video object segmentation, image saliency detection, and object segmentation benchmarks, achieving results on par with and, in many cases surpassing state-of-the-art methods. We also demonstrate the transferability of our approach to novel domains through a qualitative study on in-the-wild images. Additionally, we present extensive ablation analysis to support our design choices and highlight the contribution of each component of our proposed method.

Abstract: Many real-world applications, from sport analysis to surveillance, benefit from automatic long-term action recognition. In the current deep learning paradigm for automatic action recognition, it is imperative that models are trained and tested on datasets and tasks that evaluate if such models actually learn and reason over long-term information. In this work, we propose a method to evaluate how suitable a video dataset is to evaluate models for long-term action recognition. To this end, we define a long-term action as excluding all the videos that can be correctly recognized using solely short-term information. We test this definition on existing long-term classification tasks on three popular real-world datasets, namely Breakfast, CrossTask and LVU, to determine if these datasets are truly evaluating long-term recognition. Our study reveals that these datasets can be effectively solved using shortcuts based on short-term information. Following this finding, we encourage long-term action recognition researchers to make use of datasets that need long-term information to be solved.

Abstract: Previous work on long-term video action recognition relies on deep 3D-convolutional models that have a large temporal receptive field (RF). We argue that these models are not always the best choice for temporal modeling in videos. A large temporal receptive field allows the model to encode the exact sub-action order of a video, which causes a performance decrease when testing videos have a different sub-action order. In this work, we investigate whether we can improve the model robustness to the sub-action order by shrinking the temporal receptive field of action recognition models. For this, we design Video BagNet, a variant of the 3D ResNet-50 model with the temporal receptive field size limited to 1, 9, 17 or 33 frames. We analyze Video BagNet on synthetic and real-world video datasets and experimentally compare models with varying temporal receptive fields. We find that short receptive fields are robust to sub-action order changes, while larger temporal receptive fields are sensitive to the sub-action order.

Abstract: Generating both plausible and accurate full body avatar motion is the key to the quality of immersive experiences in mixed reality scenarios. Head-Mounted Devices (HMDs) typically only provide a few input signals, such as head and hands 6-DoF. Recently, different approaches achieved impressive performance in generating full body motion given only head and hands signal. However, to the best of our knowledge, all existing approaches rely on full hand visibility. While this is the case when, e.g., using motion controllers, a considerable proportion of mixed reality experiences do not involve motion controllers and instead rely on egocentric hand tracking. This introduces the challenge of partial hand visibility owing to the restricted field of view of the HMD. In this paper, we propose the first unified approach, HMD-NeMo, that addresses plausible and accurate full body motion generation even when the hands may be only partially visible. HMD-NeMo is a lightweight neural network that predicts the full body motion in an online and real-time fashion. At the heart of HMD-NeMo is the spatio-temporal encoder with novel temporally adaptable mask tokens that encourage plausible motion in the absence of hand observations. We perform extensive analysis of the impact of different components in HMD-NeMo and introduce a new state-of-the-art on AMASS dataset through our evaluation.

Abstract: Motivated by the challenges of the Digital Ancient Near Eastern Studies (DANES) community, we develop digital tools for processing cuneiform script being a 3D script imprinted into clay tablets used for more than three millennia and at least eight major languages. It consists of thousands of characters that have changed over time and space. Photographs are the most common representations usable for machine learning, while ink drawings are prone to interpretation. Best suited 3D datasets that are becoming available. We created and used the HeiCuBeDa and MaiCuBeDa datasets, which consist of around 500 annotated tablets. For our novel OCR-like approach to mixed image data, we provide an additional mapping tool for transferring annotations between 3D renderings and photographs. Our sign localization uses a RepPoints detector to predict the locations of characters as bounding boxes. We use image data from GigaMesh's MSII (curvature, see https://gigamesh.eu) based rendering, Phong-shaded 3D models, and photographs as well as illumination augmentation. The results show that using rendered 3D images for sign detection performs better than other work on photographs. In addition, our approach gives reasonably good results for photographs only, while it is best used for mixed datasets. More importantly, the Phong renderings, and especially the MSII renderings, improve the results on photographs, which is the largest dataset on a global scale.

Abstract: The quality of a wood log in the wood industry depends heavily on the presence of both outer and inner defects, including inner knots that are a result of the growth of tree branches. Today, locating the inner knots require the use of expensive equipment such as X-ray scanners. In this paper, we address the task of predicting the location of inner defects from the outer shape of the logs. The dataset is built by extracting both the contours and the knots with X-ray measurements. We propose to solve this binary segmentation task by leveraging convolutional recurrent neural networks. Once the neural network is trained, inference can be performed from the outer shape measured with cheap devices such as laser profilers. We demonstrate the effectiveness of our approach on fir and spruce tree species and perform ablation on the recurrence to demonstrate its importance.

Abstract: Intracranial hemorrhage (ICH) is a pathological condition characterized by bleeding inside the skull or brain, which can be attributed to various factors. Identifying, localizing and quantifying ICH has important clinical implications, in a bleed-dependent manner. While deep learning techniques are widely used in medical image segmentation and have been applied to the ICH segmentation task, existing public ICH datasets do not support the multi-class segmentation problem. To address this, we develop the Brain Hemorrhage Segmentation Dataset (BHSD), which provides a 3D multi-class ICH dataset containing 192 volumes with pixel-level annotations and 2200 volumes with slice-level annotations across five categories of ICH. To demonstrate the utility of the dataset, we formulate a series of supervised and semi-supervised ICH segmentation tasks. We provide experimental results with state-of-the-art models as reference benchmarks for further model developments and evaluations on this dataset.

Abstract: Humans effortlessly infer the 3D shape of objects. What computations underlie this ability? Although various computational models have been proposed, none of them capture the human ability to match object shape across viewpoints. Here, we ask whether and how this gap might be closed. We begin with a relatively novel class of computational models, 3D neural fields, which encapsulate the basic principles of classic analysis-by-synthesis in a deep neural network (DNN). First, we find that a 3D Light Field Network (3D-LFN) supports 3D matching judgments well aligned to humans for within-category comparisons, adversarially-defined comparisons that accentuate the 3D failure cases of standard DNN models, and adversarially-defined comparisons for algorithmically generated shapes with no category structure. We then investigate the source of the 3D-LFN's ability to achieve human-aligned performance through a series of computational experiments. Exposure to multiple viewpoints of objects during training and a multi-view learning objective are the primary factors behind model-human alignment; even conventional DNN architectures come much closer to human behavior when trained with multi-view objectives. Finally, we find that while the models trained with multi-view learning objectives are able to partially generalize to new object categories, they fall short of human alignment. This work provides a foundation for understanding human shape inferences within neurally mappable computational architectures and highlights important questions for future work.

Abstract: In this paper we show how Group Equivariant Convolutional Neural Networks use subsampling to learn to break equivariance to their symmetries. We focus on 2D rotations and reflections and investigate the impact of broken equivariance on network performance. We show that a change in the input dimension of a network as small as a single pixel can be enough for commonly used architectures to become approximately equivariant, rather than exactly. We investigate the impact of networks not being exactly equivariant and find that approximately equivariant networks generalise significantly worse to unseen symmetries compared to their exactly equivariant counterparts. However, when the symmetries in the training data are not identical to the symmetries of the network, we find that approximately equivariant networks are able to relax their own equivariant constraints, causing them to match or outperform exactly equivariant networks on common benchmark datasets.

Abstract: Existing visual tracking methods typically take an image patch as the reference of the target to perform tracking. However, a single image patch cannot provide a complete and precise concept of the target object as images are limited in their ability to abstract and can be ambiguous, which makes it difficult to track targets with drastic variations. In this paper, we propose the CiteTracker to enhance target modeling and inference in visual tracking by connecting images and text. Specifically, we develop a text generation module to convert the target image patch into a descriptive text containing its class and attribute information, providing a comprehensive reference point for the target. In addition, a dynamic description module is designed to adapt to target variations for more effective target representation. We then associate the target description and the search image using an attention-based correlation module to generate the correlated features for target state reference. Extensive experiments on five diverse datasets are conducted to evaluate the proposed algorithm and the favorable performance against the state-of-the-art methods demonstrates the effectiveness of the proposed tracking method.

Abstract: Current video object detection (VOD) models often encounter issues with over-aggregation due to redundant aggregation strategies, which perform feature aggregation on every frame. This results in suboptimal performance and increased computational complexity. In this work, we propose an image-level Object Detection Difficulty (ODD) metric to quantify the difficulty of detecting objects in a given image. The derived ODD scores can be used in the VOD process to mitigate over-aggregation. Specifically, we train an ODD predictor as an auxiliary head of a still-image object detector to compute the ODD score for each image based on the discrepancies between detection results and ground-truth bounding boxes. The ODD score enhances the VOD system in two ways: 1) it enables the VOD system to select superior global reference frames, thereby improving overall accuracy; and 2) it serves as an indicator in the newly designed ODD Scheduler to eliminate the aggregation of frames that are easy to detect, thus accelerating the VOD process. Comprehensive experiments demonstrate that, when utilized for selecting global reference frames, ODD-VOD consistently enhances the accuracy of Global-frame-based VOD models. When employed for acceleration, ODD-VOD consistently improves the frames per second (FPS) by an average of 73.3% across 8 different VOD models without sacrificing accuracy. When combined, ODD-VOD attains state-of-the-art performance when competing with many VOD methods in both accuracy and speed. Our work represents a significant advancement towards making VOD more practical for real-world applications.

Abstract: Cross-modal pre-training has shown impressive performance on a wide range of downstream tasks, benefiting from massive image-text pairs collected from the Internet. In practice, online data are growing constantly, highlighting the importance of the ability of pre-trained model to learn from data that is continuously growing. Existing works on cross-modal pre-training mainly focus on training a network with fixed architecture. However, it is impractical to limit the model capacity when considering the continuously growing nature of pre-training data in real-world applications. On the other hand, it is important to utilize the knowledge in the current model to obtain efficient training and better performance. To address the above issues, in this paper, we propose GrowCLIP, a data-driven automatic model growing algorithm for contrastive language-image pre-training with continuous image-text pairs as input. Specially, we adopt a dynamic growth space and seek out the optimal architecture at each growth step to adapt to online learning scenarios. And the shared encoder is proposed in our growth space to enhance the degree of cross-modal fusion. Besides, we explore the effect of growth in different dimensions, which could provide future references for the design of cross-modal model architecture. Finally, we employ parameter inheriting with momentum (PIM) to maintain the previous knowledge and address the issue of the local minimum dilemma. Compared with the existing methods, GrowCLIP improves 2.3% average top-1 accuracy on zero-shot image classification of 9 downstream tasks. As for zero-shot image retrieval, GrowCLIP can improve 1.2% for top-1 image-to-text recall on Flickr30K dataset.

Abstract: In this study, Synthetic Aperture Radar (SAR) and optical data are both considered for Earth surface classification. Specifically, the integration of Sentinel-1 (S-1) and Sentinel-2 (S-2) data is carried out through supervised Machine Learning (ML) algorithms implemented on the Google Earth Engine (GEE) platform for the classification of a particular region of interest. Achieved results demonstrate how in this case radar and optical remote detection provide complementary information, benefiting surface cover classification and generally leading to increased mapping accuracy. In addition, this paper works in the direction of proving the emerging role of GEE as an effective cloud-based tool for handling large amounts of satellite data.

Abstract: Collecting diverse sets of training images for RGB-D semantic image segmentation is not always possible. In particular, when robots need to operate in privacy-sensitive areas like homes, the collection is often limited to a small set of locations. As a consequence, the annotated images lack diversity in appearance and approaches for RGB-D semantic image segmentation tend to overfit the training data. In this paper, we thus introduce semantic RGB-D image synthesis to address this problem. It requires synthesising a realistic-looking RGB-D image for a given semantic label map. Current approaches, however, are uni-modal and cannot cope with multi-modal data. Indeed, we show that extending uni-modal approaches to multi-modal data does not perform well. In this paper, we therefore propose a generator for multi-modal data that separates modal-independent information of the semantic layout from the modal-dependent information that is needed to generate an RGB and a depth image, respectively. Furthermore, we propose a discriminator that ensures semantic consistency between the label maps and the generated images and perceptual similarity between the real and generated images. Our comprehensive experiments demonstrate that the proposed method outperforms previous uni-modal methods by a large margin and that the accuracy of an approach for RGB-D semantic segmentation can be significantly improved by mixing real and generated images during training.

Abstract: Continual Learning (CL) involves training a machine learning model in a sequential manner to learn new information while retaining previously learned tasks without the presence of previous training data. Although there has been significant interest in CL, most recent CL approaches in computer vision have focused on convolutional architectures only. However, with the recent success of vision transformers, there is a need to explore their potential for CL. Although there have been some recent CL approaches for vision transformers, they either store training instances of previous tasks or require a task identifier during test time, which can be limiting. This paper proposes a new exemplar-free approach for class/task incremental learning called ConTraCon, which does not require task-id to be explicitly present during inference and avoids the need for storing previous training instances. The proposed approach leverages the transformer architecture and involves re-weighting the key, query, and value weights of the multi-head self-attention layers of a transformer trained on a similar task. The re-weighting is done using convolution, which enables the approach to maintain low parameter requirements per task. Additionally, an image augmentation-based entropic task identification approach is used to predict tasks without requiring task-ids during inference. Experiments on four benchmark datasets demonstrate that the proposed approach outperforms several competitive approaches while requiring fewer parameters.

Abstract: Modeling long-term context in videos is crucial for many fine-grained tasks including temporal action segmentation. An interesting question that is still open is how much long-term temporal context is needed for optimal performance. While transformers can model the long-term context of a video, this becomes computationally prohibitive for long videos. Recent works on temporal action segmentation thus combine temporal convolutional networks with self-attentions that are computed only for a local temporal window. While these approaches show good results, their performance is limited by their inability to capture the full context of a video. In this work, we try to answer how much long-term temporal context is required for temporal action segmentation by introducing a transformer-based model that leverages sparse attention to capture the full context of a video. We compare our model with the current state of the art on three datasets for temporal action segmentation, namely 50Salads, Breakfast, and Assembly101. Our experiments show that modeling the full context of a video is necessary to obtain the best performance for temporal action segmentation.

Abstract: Super-resolution (SR) networks have been investigated for a while, with their mobile and lightweight versions gaining noticeable popularity recently. Quantization, the procedure of decreasing the precision of network parameters (mostly FP32 to INT8), is also utilized in SR networks for establishing mobile compatibility. This study focuses on a very important but mostly overlooked post-training quantization (PTQ) step: representative dataset (RD), which adjusts the quantization range for PTQ. We propose a novel pipeline (clip-free quantization pipeline, CFQP) backed up with extensive experimental justifications to cleverly augment RD images by only using outputs of the FP32 model. Using the proposed pipeline for RD, we can successfully eliminate unwanted clipped activation layers, which nearly all mobile SR methods utilize to make the model more robust to PTQ in return for a large overhead in runtime. Removing clipped activations with our method significantly benefits overall increased stability, decreased inference runtime up to 54% on some SR models, better visual quality results compared to INT8 clipped models - and outperforms even some FP32 non-quantized models, both in runtime and visual quality, without the need for retraining with clipped activation.

Abstract: Object-centric representations using slots have shown the advances towards efficient, flexible and interpretable abstraction from low-level perceptual features in a compositional scene. Current approaches randomize the initial state of slots followed by an iterative refinement. As we show in this paper, the random slot initialization significantly affects the accuracy of the final slot prediction. Moreover, current approaches require a predetermined number of slots from prior knowledge of the data, which limits the applicability in the real world. In our work, we initialize the slot representations with clustering algorithms conditioned on the perceptual input features. This requires an additional layer in the architecture to initialize the slots given the identified clusters. We design permutation invariant and permutation equivariant versions of this layer to enable the exchangeable slot representations after clustering. Additionally, we employ mean-shift clustering to automatically identify the number of slots for a given scene. We evaluate our method on object discovery and novel view synthesis tasks with various datasets. The results show that our method outperforms prior works consistently, especially for complex scenes.

Abstract: We propose Boundary-RL, a novel weakly supervised segmentation method that utilises only patch-level labels for training. We envision the segmentation as a boundary detection problem, rather than a pixel-level classification as in previous works. This outlook on segmentation may allow for boundary delineation under challenging scenarios such as where noise artefacts may be present within the region-of-interest (ROI) boundaries, where traditional pixel-level classification-based weakly supervised methods may not be able to effectively segment the ROI. Particularly of interest, ultrasound images, where intensity values represent acoustic impedance differences between boundaries, may also benefit from the boundary delineation approach. Our method uses reinforcement learning to train a controller function to localise boundaries of ROIs using a reward derived from a pre-trained boundary-presence classifier. The classifier indicates when an object boundary is encountered within a patch, as the controller modifies the patch location in a sequential Markov decision process. The classifier itself is trained using only binary patch-level labels of object presence, which are the only labels used during training of the entire boundary delineation framework, and serves as a weak signal to inform the boundary delineation. The use of a controller function ensures that a sliding window over the entire image is not necessary. It also prevents possible false-positive or -negative cases by minimising number of patches passed to the boundary-presence classifier. We evaluate our proposed approach for a clinically relevant task of prostate gland segmentation on trans-rectal ultrasound images. We show improved performance compared to other tested weakly supervised methods, using the same labels e.g., multiple instance learning.

Abstract: Rail detection is one of the key factors for intelligent train. In the paper, motivated by the anchor line-based lane detection methods, we propose a rail detection network called DALNet based on dynamic anchor line. Aiming to solve the problem that the predefined anchor line is image agnostic, we design a novel dynamic anchor line mechanism. It utilizes a dynamic anchor line generator to dynamically generate an appropriate anchor line for each rail instance based on the position and shape of the rails in the input image. These dynamically generated anchor lines can be considered as better position references to accurately localize the rails than the predefined anchor lines. In addition, we present a challenging urban rail detection dataset DL-Rail with high-quality annotations and scenario diversity. DL-Rail contains 7000 pairs of images and annotations along with scene tags, and it is expected to encourage the development of rail detection. We extensively compare DALNet with many competitive lane methods. The results show that our DALNet achieves state-of-the-art performance on our DL-Rail rail detection dataset and the popular Tusimple and LLAMAS lane detection benchmarks. The code will be released at \url{https://github.com/Yzichen/mmLaneDet}.

Abstract: Creating high quality and realistic materials in computer graphics is a challenging and time-consuming task, which requires great expertise. In this paper, we present MatFuse, a novel unified approach that harnesses the generative power of diffusion models (DM) to simplify the creation of SVBRDF maps. Our DM-based pipeline integrates multiple sources of conditioning, such as color palettes, sketches, and pictures, enabling fine-grained control and flexibility in material synthesis. This design allows for the combination of diverse information sources (e.g., sketch + image embedding), enhancing creative possibilities in line with the principle of compositionality. We demonstrate the generative capabilities of the proposed method under various conditioning settings; on the SVBRDF estimation task, we show that our method yields performance comparable to state-of-the-art approaches, both qualitatively and quantitatively.

Abstract: We present ScanNet++, a large-scale dataset that couples together capture of high-quality and commodity-level geometry and color of indoor scenes. Each scene is captured with a high-end laser scanner at sub-millimeter resolution, along with registered 33-megapixel images from a DSLR camera, and RGB-D streams from an iPhone. Scene reconstructions are further annotated with an open vocabulary of semantics, with label-ambiguous scenarios explicitly annotated for comprehensive semantic understanding. ScanNet++ enables a new real-world benchmark for novel view synthesis, both from high-quality RGB capture, and importantly also from commodity-level images, in addition to a new benchmark for 3D semantic scene understanding that comprehensively encapsulates diverse and ambiguous semantic labeling scenarios. Currently, ScanNet++ contains 460 scenes, 280,000 captured DSLR images, and over 3.7M iPhone RGBD frames.

Abstract: Vision transformers have shown unprecedented levels of performance in tackling various visual perception tasks in recent years. However, the architectural and computational complexity of such network architectures have made them challenging to deploy in real-world applications with high-throughput, low-memory requirements. As such, there has been significant research recently on the design of efficient vision transformer architectures. In this study, we explore the generation of fast vision transformer architecture designs via generative architecture search (GAS) to achieve a strong balance between accuracy and architectural and computational efficiency. Through this generative architecture search process, we create TurboViT, a highly efficient hierarchical vision transformer architecture design that is generated around mask unit attention and Q-pooling design patterns. The resulting TurboViT architecture design achieves significantly lower architectural computational complexity (>2.47$\times$ smaller than FasterViT-0 while achieving same accuracy) and computational complexity (>3.4$\times$ fewer FLOPs and 0.9% higher accuracy than MobileViT2-2.0) when compared to 10 other state-of-the-art efficient vision transformer network architecture designs within a similar range of accuracy on the ImageNet-1K dataset. Furthermore, TurboViT demonstrated strong inference latency and throughput in both low-latency and batch processing scenarios (>3.21$\times$ lower latency and >3.18$\times$ higher throughput compared to FasterViT-0 for low-latency scenario). These promising results demonstrate the efficacy of leveraging generative architecture search for generating efficient transformer architecture designs for high-throughput scenarios.

Abstract: Existing kinematic skeleton-based 3D human pose estimation methods only predict joint positions. Although this is sufficient to compute the yaw and pitch of the bone rotations, the roll around the axis of the bones remains unresolved by these methods. In this paper, we propose a novel 2D-to-3D lifting Graph Convolution Network named PoseGraphNet++ to predict the complete human pose including the joint positions and the bone orientations. We employ node and edge convolutions to utilize the joint and bone features. Our model is evaluated on multiple benchmark datasets, and its performance is either on par with or better than the state-of-the-art in terms of both position and rotation metrics. Through extensive ablation studies, we show that PoseGraphNet++ benefits from exploiting the mutual relationship between the joints and the bones.

Abstract: Latest methods represent shapes with open surfaces using unsigned distance functions (UDFs). They train neural networks to learn UDFs and reconstruct surfaces with the gradients around the zero level set of the UDF. However, the differential networks struggle from learning the zero level set where the UDF is not differentiable, which leads to large errors on unsigned distances and gradients around the zero level set, resulting in highly fragmented and discontinuous surfaces. To resolve this problem, we propose to learn a more continuous zero level set in UDFs with level set projections. Our insight is to guide the learning of zero level set using the rest non-zero level sets via a projection procedure. Our idea is inspired from the observations that the non-zero level sets are much smoother and more continuous than the zero level set. We pull the non-zero level sets onto the zero level set with gradient constraints which align gradients over different level sets and correct unsigned distance errors on the zero level set, leading to a smoother and more continuous unsigned distance field. We conduct comprehensive experiments in surface reconstruction for point clouds, real scans or depth maps, and further explore the performance in unsupervised point cloud upsampling and unsupervised point normal estimation with the learned UDF, which demonstrate our non-trivial improvements over the state-of-the-art methods. Code is available at https://github.com/junshengzhou/LevelSetUDF .

Abstract: Face Recognition Systems (FRS) are vulnerable to morph attacks. A face morph is created by combining multiple identities with the intention to fool FRS and making it match the morph with multiple identities. Current Morph Attack Detection (MAD) can detect the morph but are unable to recover the identities used to create the morph with satisfactory outcomes. Existing work in de-morphing is mostly reference-based, i.e. they require the availability of one identity to recover the other. Sudipta et al. \cite{ref9} proposed a reference-free de-morphing technique but the visual realism of outputs produced were feeble. In this work, we propose SDeMorph (Stably Diffused De-morpher), a novel de-morphing method that is reference-free and recovers the identities of bona fides. Our method produces feature-rich outputs that are of significantly high quality in terms of definition and facial fidelity. Our method utilizes Denoising Diffusion Probabilistic Models (DDPM) by destroying the input morphed signal and then reconstructing it back using a branched-UNet. Experiments on ASML, FRLL-FaceMorph, FRLL-MorDIFF, and SMDD datasets support the effectiveness of the proposed method.

Abstract: Fast Adversarial Training (FAT) not only improves the model robustness but also reduces the training cost of standard adversarial training. However, fast adversarial training often suffers from Catastrophic Overfitting (CO), which results in poor robustness performance. Catastrophic Overfitting describes the phenomenon of a sudden and significant decrease in robust accuracy during the training of fast adversarial training. Many effective techniques have been developed to prevent Catastrophic Overfitting and improve the model robustness from different perspectives. However, these techniques adopt inconsistent training settings and require different training costs, i.e, training time and memory costs, leading to unfair comparisons. In this paper, we conduct a comprehensive study of over 10 fast adversarial training methods in terms of adversarial robustness and training costs. We revisit the effectiveness and efficiency of fast adversarial training techniques in preventing Catastrophic Overfitting from the perspective of model local nonlinearity and propose an effective Lipschitz regularization method for fast adversarial training. Furthermore, we explore the effect of data augmentation and weight averaging in fast adversarial training and propose a simple yet effective auto weight averaging method to improve robustness further. By assembling these techniques, we propose a FGSM-based fast adversarial training method equipped with Lipschitz regularization and Auto Weight averaging, abbreviated as FGSM-LAW. Experimental evaluations on four benchmark databases demonstrate the superiority of the proposed method over state-of-the-art fast adversarial training methods and the advanced standard adversarial training methods.

Abstract: Self-supervised pretraining (SSP) has emerged as a popular technique in machine learning, enabling the extraction of meaningful feature representations without labelled data. In the realm of computer vision, pretrained vision transformers (ViTs) have played a pivotal role in advancing transfer learning. Nonetheless, the escalating cost of finetuning these large models has posed a challenge due to the explosion of model size. This study endeavours to evaluate the effectiveness of pure self-supervised learning (SSL) techniques in computer vision tasks, obviating the need for finetuning, with the intention of emulating human-like capabilities in generalisation and recognition of unseen objects. To this end, we propose an evaluation protocol for zero-shot segmentation based on a prompting patch. Given a point on the target object as a prompt, the algorithm calculates the similarity map between the selected patch and other patches, upon that, a simple thresholding is applied to segment the target. Another evaluation is intra-object and inter-object similarity to gauge discriminatory ability of SSP ViTs. Insights from zero-shot segmentation from prompting and discriminatory abilities of SSP led to the design of a simple SSP approach, termed MMC. This approaches combines Masked image modelling for encouraging similarity of local features, Momentum based self-distillation for transferring semantics from global to local features, and global Contrast for promoting semantics of global features, to enhance discriminative representations of SSP ViTs. Consequently, our proposed method significantly reduces the overlap of intra-object and inter-object similarities, thereby facilitating effective object segmentation within an image. Our experiments reveal that MMC delivers top-tier results in zero-shot semantic segmentation across various datasets.

Abstract: Maintaining high efficiency and high precision are two fundamental challenges in UAV tracking due to the constraints of computing resources, battery capacity, and UAV maximum load. Discriminative correlation filters (DCF)-based trackers can yield high efficiency on a single CPU but with inferior precision. Lightweight Deep learning (DL)-based trackers can achieve a good balance between efficiency and precision but performance gains are limited by the compression rate. High compression rate often leads to poor discriminative representations. To this end, this paper aims to enhance the discriminative power of feature representations from a new feature-learning perspective. Specifically, we attempt to learn more disciminative representations with contrastive instances for UAV tracking in a simple yet effective manner, which not only requires no manual annotations but also allows for developing and deploying a lightweight model. We are the first to explore contrastive learning for UAV tracking. Extensive experiments on four UAV benchmarks, including UAV123@10fps, DTB70, UAVDT and VisDrone2018, show that the proposed DRCI tracker significantly outperforms state-of-the-art UAV tracking methods.

Abstract: The demand for accurate food quantification has increased in the recent years, driven by the needs of applications in dietary monitoring. At the same time, computer vision approaches have exhibited great potential in automating tasks within the food domain. Traditionally, the development of machine learning models for these problems relies on training data sets with pixel-level class annotations. However, this approach introduces challenges arising from data collection and ground truth generation that quickly become costly and error-prone since they must be performed in multiple settings and for thousands of classes. To overcome these challenges, the paper presents a weakly supervised methodology for training food image classification and semantic segmentation models without relying on pixel-level annotations. The proposed methodology is based on a multiple instance learning approach in combination with an attention-based mechanism. At test time, the models are used for classification and, concurrently, the attention mechanism generates semantic heat maps which are used for food class segmentation. In the paper, we conduct experiments on two meta-classes within the FoodSeg103 data set to verify the feasibility of the proposed approach and we explore the functioning properties of the attention mechanism.

Abstract: The aim of this work is to perform a multitemporal analysis using the Google Earth Engine (GEE) platform for the detection of changes in urban areas using optical data and specific machine learning (ML) algorithms. As a case study, Cairo City has been identified, in Egypt country, as one of the five most populous megacities of the last decade in the world. Classification and change detection analysis of the region of interest (ROI) have been carried out from July 2013 to July 2021. Results demonstrate the validity of the proposed method in identifying changed and unchanged urban areas over the selected period. Furthermore, this work aims to evidence the growing significance of GEE as an efficient cloud-based solution for managing large quantities of satellite data.

Abstract: Recent strides in Text-to-3D techniques have been propelled by distilling knowledge from powerful large text-to-image diffusion models (LDMs). Nonetheless, existing Text-to-3D approaches often grapple with challenges such as over-saturation, inadequate detailing, and unrealistic outputs. This study presents a novel strategy that leverages explicitly synthesized multi-view images to address these issues. Our approach involves the utilization of image-to-image pipelines, empowered by LDMs, to generate posed high-quality images based on the renderings of coarse 3D models. Although the generated images mostly alleviate the aforementioned issues, challenges such as view inconsistency and significant content variance persist due to the inherent generative nature of large diffusion models, posing extensive difficulties in leveraging these images effectively. To overcome this hurdle, we advocate integrating a discriminator alongside a novel Diffusion-GAN dual training strategy to guide the training of 3D models. For the incorporated discriminator, the synthesized multi-view images are considered real data, while the renderings of the optimized 3D models function as fake data. We conduct a comprehensive set of experiments that demonstrate the effectiveness of our method over baseline approaches.

Abstract: Video-based ambient monitoring of gait for older adults with dementia has the potential to detect negative changes in health and allow clinicians and caregivers to intervene early to prevent falls or hospitalizations. Computer vision-based pose tracking models can process video data automatically and extract joint locations; however, publicly available models are not optimized for gait analysis on older adults or clinical populations. In this work we train a deep neural network to map from a two dimensional pose sequence, extracted from a video of an individual walking down a hallway toward a wall-mounted camera, to a set of three-dimensional spatiotemporal gait features averaged over the walking sequence. The data of individuals with dementia used in this work was captured at two sites using a wall-mounted system to collect the video and depth information used to train and evaluate our model. Our Pose2Gait model is able to extract velocity and step length values from the video that are correlated with the features from the depth camera, with Spearman's correlation coefficients of .83 and .60 respectively, showing that three dimensional spatiotemporal features can be predicted from monocular video. Future work remains to improve the accuracy of other features, such as step time and step width, and test the utility of the predicted values for detecting meaningful changes in gait during longitudinal ambient monitoring.

Abstract: Given a query composed of a reference image and a relative caption, the Composed Image Retrieval goal is to retrieve images visually similar to the reference one that integrates the modifications expressed by the caption. Given that recent research has demonstrated the efficacy of large-scale vision and language pre-trained (VLP) models in various tasks, we rely on features from the OpenAI CLIP model to tackle the considered task. We initially perform a task-oriented fine-tuning of both CLIP encoders using the element-wise sum of visual and textual features. Then, in the second stage, we train a Combiner network that learns to combine the image-text features integrating the bimodal information and providing combined features used to perform the retrieval. We use contrastive learning in both stages of training. Starting from the bare CLIP features as a baseline, experimental results show that the task-oriented fine-tuning and the carefully crafted Combiner network are highly effective and outperform more complex state-of-the-art approaches on FashionIQ and CIRR, two popular and challenging datasets for composed image retrieval. Code and pre-trained models are available at https://github.com/ABaldrati/CLIP4Cir

Abstract: Gait recognition is to seek correct matches for query individuals by their unique walking patterns at a long distance. However, current methods focus solely on individual gait features, disregarding inter-personal relationships. In this paper, we reconsider gait representation, asserting that gait is not just an aggregation of individual features, but also the relationships among different subjects' gait features once reference gaits are established. From this perspective, we redefine classifier weights as reference-anchored gaits, allowing each person's gait to be described by their relationship with these references. In our work, we call this novel descriptor Relationship Descriptor (RD). This Relationship Descriptor offers two benefits: emphasizing meaningful features and enhancing robustness. To be specific, The normalized dot product between gait features and classifier weights signifies a similarity relation, where each dimension indicates the similarity between the test sample and each training ID's gait prototype, respectively. Despite its potential, the direct use of relationship descriptors poses dimensionality challenges since the dimension of RD depends on the training set's identity count. To address this, we propose a Farthest Anchored gaits Selection algorithm and a dimension reduction method to boost gait recognition performance. Our method can be built on top of off-the-shelf pre-trained classification-based models without extra parameters. We show that RD achieves higher recognition performance than directly using extracted features. We evaluate the effectiveness of our method on the popular GREW, Gait3D, CASIA-B, and OU-MVLP, showing that our method consistently outperforms the baselines and achieves state-of-the-art performances.

Abstract: Human actions in egocentric videos are often hand-object interactions composed from a verb (performed by the hand) applied to an object. Despite their extensive scaling up, egocentric datasets still face two limitations - sparsity of action compositions and a closed set of interacting objects. This paper proposes a novel open vocabulary action recognition task. Given a set of verbs and objects observed during training, the goal is to generalize the verbs to an open vocabulary of actions with seen and novel objects. To this end, we decouple the verb and object predictions via an object-agnostic verb encoder and a prompt-based object encoder. The prompting leverages CLIP representations to predict an open vocabulary of interacting objects. We create open vocabulary benchmarks on the EPIC-KITCHENS-100 and Assembly101 datasets; whereas closed-action methods fail to generalize, our proposed method is effective. In addition, our object encoder significantly outperforms existing open-vocabulary visual recognition methods in recognizing novel interacting objects.

Abstract: We are concerned with a challenging scenario in unpaired multiview video learning. In this case, the model aims to learn comprehensive multiview representations while the cross-view semantic information exhibits variations. We propose Semantics-based Unpaired Multiview Learning (SUM-L) to tackle this unpaired multiview learning problem. The key idea is to build cross-view pseudo-pairs and do view-invariant alignment by leveraging the semantic information of videos. To facilitate the data efficiency of multiview learning, we further perform video-text alignment for first-person and third-person videos, to fully leverage the semantic knowledge to improve video representations. Extensive experiments on multiple benchmark datasets verify the effectiveness of our framework. Our method also outperforms multiple existing view-alignment methods, under the more challenging scenario than typical paired or unpaired multimodal or multiview learning. Our code is available at https://github.com/wqtwjt1996/SUM-L.

Abstract: This paper studies co-segmenting the common semantic object in a set of images. Existing works either rely on carefully engineered networks to mine the implicit semantic information in visual features or require extra data (i.e., classification labels) for training. In this paper, we leverage the contrastive language-image pre-training framework (CLIP) for the task. With a backbone segmentation network that independently processes each image from the set, we introduce semantics from CLIP into the backbone features, refining them in a coarse-to-fine manner with three key modules: i) an image set feature correspondence module, encoding global consistent semantic information of the image set; ii) a CLIP interaction module, using CLIP-mined common semantics of the image set to refine the backbone feature; iii) a CLIP regularization module, drawing CLIP towards this co-segmentation task, identifying the best CLIP semantic and using it to regularize the backbone feature. Experiments on four standard co-segmentation benchmark datasets show that the performance of our method outperforms state-of-the-art methods.

Abstract: Recently, large-scale pre-trained vision-language models (e.g. CLIP and ALIGN) have demonstrated remarkable effectiveness in acquiring transferable visual representations. To leverage the valuable knowledge encoded within these models for downstream tasks, several fine-tuning approaches, including prompt tuning methods and adapter-based methods, have been developed to adapt vision-language models effectively with supervision. However, these methods rely on the availability of annotated samples, which can be labor-intensive and time-consuming to acquire, thus limiting scalability. To address this issue, in this work, we design an unsupervised fine-tuning approach for vision-language models called Unsupervised Prototype Adapter (UP-Adapter). Specifically, for the unannotated target datasets, we leverage the text-image aligning capability of CLIP to automatically select the most confident samples for each class. Utilizing these selected samples, we generate class prototypes, which serve as the initialization for the learnable prototype model. After fine-tuning, the prototype model prediction is combined with the original CLIP's prediction by a residual connection to perform downstream recognition tasks. Our extensive experimental results on image recognition and domain generalization show that the proposed unsupervised method outperforms 8-shot CoOp, 8-shot Tip-Adapter, and also the state-of-the-art UPL method by large margins.

Abstract: In recent years, vision transformers have been introduced into face recognition and analysis and have achieved performance breakthroughs. However, most previous methods generally train a single model or an ensemble of models to perform the desired task, which ignores the synergy among different tasks and fails to achieve improved prediction accuracy, increased data efficiency, and reduced training time. This paper presents a multi-purpose algorithm for simultaneous face recognition, facial expression recognition, age estimation, and face attribute estimation (40 attributes including gender) based on a single Swin Transformer. Our design, the SwinFace, consists of a single shared backbone together with a subnet for each set of related tasks. To address the conflicts among multiple tasks and meet the different demands of tasks, a Multi-Level Channel Attention (MLCA) module is integrated into each task-specific analysis subnet, which can adaptively select the features from optimal levels and channels to perform the desired tasks. Extensive experiments show that the proposed model has a better understanding of the face and achieves excellent performance for all tasks. Especially, it achieves 90.97% accuracy on RAF-DB and 0.22 $\epsilon$-error on CLAP2015, which are state-of-the-art results on facial expression recognition and age estimation respectively. The code and models will be made publicly available at https://github.com/lxq1000/SwinFace.

Abstract: The field of multi-object tracking has recently seen a renewed interest in the good old schema of tracking-by-detection, as its simplicity and strong priors spare it from the complex design and painful babysitting of tracking-by-attention approaches. In view of this, we aim at extending tracking-by-detection to multi-modal settings, where a comprehensive cost has to be computed from heterogeneous information e.g., 2D motion cues, visual appearance, and pose estimates. More precisely, we follow a case study where a rough estimate of 3D information is also available and must be merged with other traditional metrics (e.g., the IoU). To achieve that, recent approaches resort to either simple rules or complex heuristics to balance the contribution of each cost. However, i) they require careful tuning of tailored hyperparameters on a hold-out set, and ii) they imply these costs to be independent, which does not hold in reality. We address these issues by building upon an elegant probabilistic formulation, which considers the cost of a candidate association as the negative log-likelihood yielded by a deep density estimator, trained to model the conditional joint probability distribution of correct associations. Our experiments, conducted on both simulated and real benchmarks, show that our approach consistently enhances the performance of several tracking-by-detection algorithms.

Abstract: Video-Text Retrieval (VTR) is a crucial multi-modal task in an era of massive video-text data on the Internet. A plethora of work characterized by using a two-stream Vision-Language model architecture that learns a joint representation of video-text pairs has become a prominent approach for the VTR task. However, these models operate under the assumption of bijective video-text correspondences and neglect a more practical scenario where video content usually encompasses multiple events, while texts like user queries or webpage metadata tend to be specific and correspond to single events. This establishes a gap between the previous training objective and real-world applications, leading to the potential performance degradation of earlier models during inference. In this study, we introduce the Multi-event Video-Text Retrieval (MeVTR) task, addressing scenarios in which each video contains multiple different events, as a niche scenario of the conventional Video-Text Retrieval Task. We present a simple model, Me-Retriever, which incorporates key event video representation and a new MeVTR loss for the MeVTR task. Comprehensive experiments show that this straightforward framework outperforms other models in the Video-to-Text and Text-to-Video tasks, effectively establishing a robust baseline for the MeVTR task. We believe this work serves as a strong foundation for future studies. Code is available at https://github.com/gengyuanmax/MeVTR.

Abstract: This report details the method of the winning entry of the AVDN Challenge in ICCV 2023. The competition addresses the Aerial Navigation from Dialog History (ANDH) task, which requires a drone agent to associate dialog history with aerial observations to reach the destination. For better cross-modal grounding abilities of the drone agent, we propose a Target-Grounded Graph-Aware Transformer (TG-GAT) framework. Concretely, TG-GAT first leverages a graph-aware transformer to capture spatiotemporal dependency, which is beneficial for navigation state tracking and robust action planning. TG-GAT first leverages a graph-aware transformer to capture spatiotemporal dependencies for more robust action planning. In addition, an auxiliary visual grounding task is devised to boost the agent's awareness of referred landmarks. Moreover, a hybrid augmentation strategy based on large language models is utilized to mitigate data scarcity limitations. Our TG-GAT framework won the AVDN Challenge 2023, with 2.2% and 3.0% absolute improvements over the baseline on SPL and SR metrics, respectively. The code is available at https://github.com/yifeisu/avdn-challenge.

Abstract: Recent studies show that self-attentions behave like low-pass filters (as opposed to convolutions) and enhancing their high-pass filtering capability improves model performance. Contrary to this idea, we investigate existing convolution-based models with spectral analysis and observe that improving the low-pass filtering in convolution operations also leads to performance improvement. To account for this observation, we hypothesize that utilizing optimal token mixers that capture balanced representations of both high- and low-frequency components can enhance the performance of models. We verify this by decomposing visual features into the frequency domain and combining them in a balanced manner. To handle this, we replace the balancing problem with a mask filtering problem in the frequency domain. Then, we introduce a novel token-mixer named SPAM and leverage it to derive a MetaFormer model termed as SPANet. Experimental results show that the proposed method provides a way to achieve this balance, and the balanced representations of both high- and low-frequency components can improve the performance of models on multiple computer vision tasks. Our code is available at $\href{https://doranlyong.github.io/projects/spanet/}{\text{https://doranlyong.github.io/projects/spanet/}}$.

Abstract: This study introduces a novel framework, G3Reg, for fast and robust global registration of LiDAR point clouds. In contrast to conventional complex keypoints and descriptors, we extract fundamental geometric primitives including planes, clusters, and lines (PCL) from the raw point cloud to obtain low-level semantic segments. Each segment is formulated as a unified Gaussian Ellipsoid Model (GEM) by employing a probability ellipsoid to ensure the ground truth centers are encompassed with a certain degree of probability. Utilizing these GEMs, we then present a distrust-and-verify scheme based on a Pyramid Compatibility Graph for Global Registration (PAGOR). Specifically, we establish an upper bound, which can be traversed based on the confidence level for compatibility testing to construct the pyramid graph. Gradually, we solve multiple maximum cliques (MAC) for each level of the graph, generating numerous transformation candidates. In the verification phase, we adopt a precise and efficient metric for point cloud alignment quality, founded on geometric primitives, to identify the optimal candidate. The performance of the algorithm is extensively validated on three publicly available datasets and a self-collected multi-session dataset, without changing any parameter settings in the experimental evaluation. The results exhibit superior robustness and real-time performance of the G3Reg framework compared to state-of-the-art methods. Furthermore, we demonstrate the potential for integrating individual GEM and PAGOR components into other algorithmic frameworks to enhance their efficacy. To advance further research and promote community understanding, we have publicly shared the source code.

Abstract: Large-scale foundation models, such as CLIP, have demonstrated remarkable success in visual recognition tasks by embedding images in a semantically rich space. Self-supervised learning (SSL) has also shown promise in improving visual recognition by learning invariant features. However, the combination of CLIP with SSL is found to face challenges due to the multi-task framework that blends CLIP's contrastive loss and SSL's loss, including difficulties with loss weighting and inconsistency among different views of images in CLIP's output space. To overcome these challenges, we propose a prompt learning-based model called GOPro, which is a unified framework that ensures similarity between various augmented views of input images in a shared image-text embedding space, using a pair of learnable image and text projectors atop CLIP, to promote invariance and generalizability. To automatically learn such prompts, we leverage the visual content and style primitives extracted from pre-trained CLIP and adapt them to the target task. In addition to CLIP's cross-domain contrastive loss, we introduce a visual contrastive loss and a novel prompt consistency loss, considering the different views of the images. GOPro is trained end-to-end on all three loss objectives, combining the strengths of CLIP and SSL in a principled manner. Empirical evaluations demonstrate that GOPro outperforms the state-of-the-art prompting techniques on three challenging domain generalization tasks across multiple benchmarks by a significant margin. Our code is available at https://github.com/mainaksingha01/GOPro.

Abstract: Generating video stories from text prompts is a complex task. In addition to having high visual quality, videos need to realistically adhere to a sequence of text prompts whilst being consistent throughout the frames. Creating a benchmark for video generation requires data annotated over time, which contrasts with the single caption used often in video datasets. To fill this gap, we collect comprehensive human annotations on three existing datasets, and introduce StoryBench: a new, challenging multi-task benchmark to reliably evaluate forthcoming text-to-video models. Our benchmark includes three video generation tasks of increasing difficulty: action execution, where the next action must be generated starting from a conditioning video; story continuation, where a sequence of actions must be executed starting from a conditioning video; and story generation, where a video must be generated from only text prompts. We evaluate small yet strong text-to-video baselines, and show the benefits of training on story-like data algorithmically generated from existing video captions. Finally, we establish guidelines for human evaluation of video stories, and reaffirm the need of better automatic metrics for video generation. StoryBench aims at encouraging future research efforts in this exciting new area.

Abstract: Recent advances of monocular 3D object detection facilitate the 3D multi-object tracking task based on low-cost camera sensors. In this paper, we find that the motion cue of objects along different time frames is critical in 3D multi-object tracking, which is less explored in existing monocular-based approaches. In this paper, we propose a motion-aware framework for monocular 3D MOT. To this end, we propose MoMA-M3T, a framework that mainly consists of three motion-aware components. First, we represent the possible movement of an object related to all object tracklets in the feature space as its motion features. Then, we further model the historical object tracklet along the time frame in a spatial-temporal perspective via a motion transformer. Finally, we propose a motion-aware matching module to associate historical object tracklets and current observations as final tracking results. We conduct extensive experiments on the nuScenes and KITTI datasets to demonstrate that our MoMA-M3T achieves competitive performance against state-of-the-art methods. Moreover, the proposed tracker is flexible and can be easily plugged into existing image-based 3D object detectors without re-training. Code and models are available at https://github.com/kuanchihhuang/MoMA-M3T.

Abstract: Hands are dexterous and highly versatile manipulators that are central to how humans interact with objects and their environment. Consequently, modeling realistic hand-object interactions, including the subtle motion of individual fingers, is critical for applications in computer graphics, computer vision, and mixed reality. Prior work on capturing and modeling humans interacting with objects in 3D focuses on the body and object motion, often ignoring hand pose. In contrast, we introduce GRIP, a learning-based method that takes, as input, the 3D motion of the body and the object, and synthesizes realistic motion for both hands before, during, and after object interaction. As a preliminary step before synthesizing the hand motion, we first use a network, ANet, to denoise the arm motion. Then, we leverage the spatio-temporal relationship between the body and the object to extract two types of novel temporal interaction cues, and use them in a two-stage inference pipeline to generate the hand motion. In the first stage, we introduce a new approach to enforce motion temporal consistency in the latent space (LTC), and generate consistent interaction motions. In the second stage, GRIP generates refined hand poses to avoid hand-object penetrations. Given sequences of noisy body and object motion, GRIP upgrades them to include hand-object interaction. Quantitative experiments and perceptual studies demonstrate that GRIP outperforms baseline methods and generalizes to unseen objects and motions from different motion-capture datasets.