Computer Vision and Pattern Recognition (cs.CV)

Abstract: Lane marker detection is a crucial component of the autonomous driving and driver assistance systems. Modern deep lane detection methods with row-based lane representation exhibit excellent performance on lane detection benchmarks. Through preliminary oracle experiments, we firstly disentangle the lane representation components to determine the direction of our approach. We show that correct lane positions are already among the predictions of an existing row-based detector, and the confidence scores that accurately represent intersection-over-union (IoU) with ground truths are the most beneficial. Based on the finding, we propose LaneIoU that better correlates with the metric, by taking the local lane angles into consideration. We develop a novel detector coined CLRerNet featuring LaneIoU for the target assignment cost and loss functions aiming at the improved quality of confidence scores. Through careful and fair benchmark including cross validation, we demonstrate that CLRerNet outperforms the state-of-the-art by a large margin - enjoying F1 score of 81.43% compared with 80.47% of the existing method on CULane, and 86.47% compared with 86.10% on CurveLanes.

Abstract: With the advance of large-scale model technologies, parameter-efficient transfer learning (PETL) has swept across various fields of Artificial Intelligence. Its core idea is to adapt the model to downstream tasks using only a small number of parameters. Recently, some studies have applied these techniques proven effective to multimodal tasks. However, two critical issues remain unresolved: how to further reduce the complexity with lightweight design and how to boost alignment between modalities under extremely low parameters. In this paper, we propose A graceful prompt framework for cross-modal transfer (Aurora) to overcome these challenges. Considering the redundancy in existing architectures, we first utilize the mode approximation to generate few trainable parameters to implement the multi-modal prompt tuning, which explores the low intrinsic dimension with only 0.05% parameters of the pre-trained model. Then, to better narrow the modality gap, we propose the informative context enhancement and gated query transformation modules under extremely few parameters scenes. A thorough evaluation of the Aurora on six cross-modal downstream benchmarks shows that it not only outperforms the state-of-the-art, but even outperforms the full fine-tuning approach. Our code is available at: https://github.com/WillDreamer/Aurora.

Abstract: Pre-training has emerged as an effective technique for learning powerful person representations. Most existing methods have shown that pre-training on pure-vision large-scale datasets like ImageNet and LUPerson has achieved remarkable performance. However, solely relying on visual information, the absence of robust explicit indicators poses a challenge for these methods to learn discriminative person representations. Drawing inspiration from the intrinsic fine-grained attribute indicators of person descriptions, we explore introducing the language modality into person representation learning. To this end, we propose a novel language-image pre-training framework for person representation learning, termed PLIP. To explicitly build fine-grained cross-modal associations, we specifically design three pretext tasks, \ie semantic-fused image colorization, visual-fused attributes prediction, and vision-language matching. In addition, due to the lack of an appropriate dataset, we present a large-scale person dataset named SYNTH-PEDES, where the Stylish Pedestrian Attributes-union Captioning method is proposed to synthesize diverse textual descriptions. We pre-train PLIP on SYNTH-PEDES and evaluate our model by spanning downstream tasks such as text-based Re-ID, image-based Re-ID, and person attribute recognition. Extensive experiments demonstrate that our model not only significantly improves existing methods on all these tasks, but also shows great ability in the few-shot and domain generalization settings. The code, dataset and weights will be released at~\url{https://github.com/Zplusdragon/PLIP}

Abstract: Automatically narrating a video with natural language can assist people in grasping and managing massive videos on the Internet. From the perspective of video uploaders, they may have varied preferences for writing the desired video description to attract more potential followers, e.g. catching customers' attention for product videos. The Controllable Video Captioning task is therefore proposed to generate a description conditioned on the user demand and video content. However, existing works suffer from two shortcomings: 1) the control signal is fixed and can only express single-grained control; 2) the video description can not be further edited to meet dynamic user demands. In this paper, we propose a novel Video Description Editing (VDEdit) task to automatically revise an existing video description guided by flexible user requests. Inspired by human writing-revision habits, we design the user command as a {operation, position, attribute} triplet to cover multi-grained use requirements, which can express coarse-grained control (e.g. expand the description) as well as fine-grained control (e.g. add specified details in specified position) in a unified format. To facilitate the VDEdit task, we first automatically construct a large-scale benchmark dataset namely VATEX-EDIT in the open domain describing diverse human activities. Considering the real-life application scenario, we further manually collect an e-commerce benchmark dataset called EMMAD-EDIT. We propose a unified framework to convert the {operation, position, attribute} triplet into a textual control sequence to handle multi-grained editing commands. For VDEdit evaluation, we adopt comprehensive metrics to measure three aspects of model performance, including caption quality, caption-command consistency, and caption-video alignment.

Abstract: In recent years, several video quality assessment (VQA) methods have been developed, achieving high performance. However, these methods were not specifically trained for enhanced videos, which limits their ability to predict video quality accurately based on human subjective perception. To address this issue, we propose a stack-based framework for VQA that outperforms existing state-of-the-art methods on VDPVE, a dataset consisting of enhanced videos. In addition to proposing the VQA framework for enhanced videos, we also investigate its application on professionally generated content (PGC). To address copyright issues with premium content, we create the PGCVQ dataset, which consists of videos from YouTube. We evaluate our proposed approach and state-of-the-art methods on PGCVQ, and provide new insights on the results. Our experiments demonstrate that existing VQA algorithms can be applied to PGC videos, and we find that VQA performance for PGC videos can be improved by considering the plot of a play, which highlights the importance of video semantic understanding.

Abstract: Earth observation (EO) is a prime instrument for monitoring land and ocean processes, studying the dynamics at work, and taking the pulse of our planet. This article gives a bird's eye view of the essential scientific tools and approaches informing and supporting the transition from raw EO data to usable EO-based information. The promises, as well as the current challenges of these developments, are highlighted under dedicated sections. Specifically, we cover the impact of (i) Computer vision; (ii) Machine learning; (iii) Advanced processing and computing; (iv) Knowledge-based AI; (v) Explainable AI and causal inference; (vi) Physics-aware models; (vii) User-centric approaches; and (viii) the much-needed discussion of ethical and societal issues related to the massive use of ML technologies in EO.

Abstract: Target tracking and trajectory modeling have important applications in surveillance video analysis and have received great attention in the fields of road safety and community security. In this work, we propose a lightweight real-time video analysis scheme that uses a model learned from motion patterns to monitor the behavior of objects, which can be used for applications such as real-time representation and prediction. The proposed sequence clustering algorithm based on discrete sequences makes the system have continuous online learning ability. The intrinsic repeatability of the target object trajectory is used to automatically construct the behavioral model in the three processes of feature extraction, cluster learning, and model application. In addition to the discretization of trajectory features and simple model applications, this paper focuses on online clustering algorithms and their incremental learning processes. Finally, through the learning of the trajectory model of the actual surveillance video image, the feasibility of the algorithm is verified. And the characteristics and performance of the clustering algorithm are discussed in the analysis. This scheme has real-time online learning and processing of motion models while avoiding a large number of arithmetic operations, which is more in line with the application scenarios of front-end intelligent perception.

Abstract: Domain adaptation is essential for activity recognition, as common spatiotemporal architectures risk overfitting due to increased parameters arising from the temporal dimension. Unsupervised domain adaptation methods have been extensively studied, yet, they require large-scale unlabeled data from the target domain. In this work, we address few-shot domain adaptation for video-based activity recognition (FSDA-AR), which leverages a very small amount of labeled target videos to achieve effective adaptation. This setting is attractive and promising for applications, as it requires recording and labeling only a few, or even a single example per class in the target domain, which often includes activities that are rare yet crucial to recognize. We construct FSDA-AR benchmarks using five established datasets: UCF101, HMDB51, EPIC-KITCHEN, Sims4Action, and Toyota Smart Home. Our results demonstrate that FSDA-AR performs comparably to unsupervised domain adaptation with significantly fewer (yet labeled) target examples. We further propose a novel approach, FeatFSDA, to better leverage the few labeled target domain samples as knowledge guidance. FeatFSDA incorporates a latent space semantic adjacency loss, a domain prototypical similarity loss, and a graph-attentive-network-based edge dropout technique. Our approach achieves state-of-the-art performance on all datasets within our FSDA-AR benchmark. To encourage future research of few-shot domain adaptation for video-based activity recognition, we will release our benchmarks and code at https://github.com/KPeng9510/FeatFSDA.

Abstract: In recent years, there has been growing concern over the vulnerability of convolutional neural networks (CNNs) to image perturbations. However, achieving general robustness against different types of perturbations remains challenging, in which enhancing robustness to some perturbations (e.g., adversarial perturbations) may degrade others (e.g., common corruptions). In this paper, we demonstrate that adversarial training with an emphasis on phase components significantly improves model performance on clean, adversarial, and common corruption accuracies. We propose a frequency-based data augmentation method, Adversarial Amplitude Swap, that swaps the amplitude spectrum between clean and adversarial images to generate two novel training images: adversarial amplitude and adversarial phase images. These images act as substitutes for adversarial images and can be implemented in various adversarial training setups. Through extensive experiments, we demonstrate that our method enables the CNNs to gain general robustness against different types of perturbations and results in a uniform performance against all types of common corruptions.

Abstract: We call on the Document AI (DocAI) community to reevaluate current methodologies and embrace the challenge of creating more practically-oriented benchmarks. Document Understanding Dataset and Evaluation (DUDE) seeks to remediate the halted research progress in understanding visually-rich documents (VRDs). We present a new dataset with novelties related to types of questions, answers, and document layouts based on multi-industry, multi-domain, and multi-page VRDs of various origins, and dates. Moreover, we are pushing the boundaries of current methods by creating multi-task and multi-domain evaluation setups that more accurately simulate real-world situations where powerful generalization and adaptation under low-resource settings are desired. DUDE aims to set a new standard as a more practical, long-standing benchmark for the community, and we hope that it will lead to future extensions and contributions that address real-world challenges. Finally, our work illustrates the importance of finding more efficient ways to model language, images, and layout in DocAI.

Abstract: Semantic segmentation has recently witnessed great progress. Despite the impressive overall results, the segmentation performance in some hard areas (e.g., small objects or thin parts) is still not promising. A straightforward solution is hard sample mining, which is widely used in object detection. Yet, most existing hard pixel mining strategies for semantic segmentation often rely on pixel's loss value, which tends to decrease during training. Intuitively, the pixel hardness for segmentation mainly depends on image structure and is expected to be stable. In this paper, we propose to learn pixel hardness for semantic segmentation, leveraging hardness information contained in global and historical loss values. More precisely, we add a gradient-independent branch for learning a hardness level (HL) map by maximizing hardness-weighted segmentation loss, which is minimized for the segmentation head. This encourages large hardness values in difficult areas, leading to appropriate and stable HL map. Despite its simplicity, the proposed method can be applied to most segmentation methods with no and marginal extra cost during inference and training, respectively. Without bells and whistles, the proposed method achieves consistent/significant improvement (1.37% mIoU on average) over most popular semantic segmentation methods on Cityscapes dataset, and demonstrates good generalization ability across domains. The source codes are available at https://github.com/Menoly-xin/Hardness-Level-Learning .

Abstract: This study introduces an efficacious approach, Masked Collaborative Contrast (MCC), to emphasize semantic regions in weakly supervised semantic segmentation. MCC adroitly incorporates concepts from masked image modeling and contrastive learning to devise Transformer blocks that induce keys to contract towards semantically pertinent regions. Unlike prevalent techniques that directly eradicate patch regions in the input image when generating masks, we scrutinize the neighborhood relations of patch tokens by exploring masks considering keys on the affinity matrix. Moreover, we generate positive and negative samples in contrastive learning by utilizing the masked local output and contrasting it with the global output. Elaborate experiments on commonly employed datasets evidences that the proposed MCC mechanism effectively aligns global and local perspectives within the image, attaining impressive performance.

Abstract: Industrial visual inspection aims at detecting surface defects in products during the manufacturing process. Although existing anomaly detection models have shown great performance on many public benchmarks, their limited adjustability and ability to detect logical anomalies hinder their broader use in real-world settings. To this end, in this paper, we propose a novel component-aware anomaly detection framework (ComAD) which can simultaneously achieve adjustable and logical anomaly detection for industrial scenarios. Specifically, we propose to segment images into multiple components based on a lightweight and nearly training-free unsupervised semantic segmentation model. Then, we design an interpretable logical anomaly detection model through modeling the metrological features of each component and their relationships. Despite its simplicity, our framework achieves state-of-the-art performance on image-level logical anomaly detection. Meanwhile, segmenting a product image into multiple components provides a novel perspective for industrial visual inspection, demonstrating great potential in model customization, noise resistance, and anomaly classification. The code will be available at https://github.com/liutongkun/ComAD.

Abstract: Deep learning-based image compression methods have led to high rate-distortion performances compared to traditional codecs. Recently, Generative Adversarial Networks (GANs)-based compression models, e.g., High Fidelity Compression (HiFiC), have attracted great attention in the computer vision community. However, most of these works aim for spatial compression only and do not consider the spatio-spectral redundancies observed in hyperspectral images (HSIs). To address this problem, in this paper, we adapt the HiFiC spatial compression model to perform spatio-spectral compression of HSIs. To this end, we introduce two new models: i) HiFiC using Squeeze and Excitation (SE) blocks (denoted as HiFiC$_{SE}$); and ii) HiFiC with 3D convolutions (denoted as HiFiC$_{3D}$). We analyze the effectiveness of HiFiC$_{SE}$ and HiFiC$_{3D}$ in exploiting the spatio-spectral redundancies with channel attention and inter-dependency analysis. Experimental results show the efficacy of the proposed models in performing spatio-spectral compression and reconstruction at reduced bitrates and higher reconstruction quality when compared to JPEG 2000 and the standard HiFiC spatial compression model. The code of the proposed models is publicly available at https://git.tu-berlin.de/rsim/HSI-SSC .

Abstract: Dynamic scene graphs generated from video clips could help enhance the semantic visual understanding in a wide range of challenging tasks such as environmental perception, autonomous navigation, and task planning of self-driving vehicles and mobile robots. In the process of temporal and spatial modeling during dynamic scene graph generation, it is particularly intractable to learn time-variant relations in dynamic scene graphs among frames. In this paper, we propose a Time-variant Relation-aware TRansformer (TR$^2$), which aims to model the temporal change of relations in dynamic scene graphs. Explicitly, we leverage the difference of text embeddings of prompted sentences about relation labels as the supervision signal for relations. In this way, cross-modality feature guidance is realized for the learning of time-variant relations. Implicitly, we design a relation feature fusion module with a transformer and an additional message token that describes the difference between adjacent frames. Extensive experiments on the Action Genome dataset prove that our TR$^2$ can effectively model the time-variant relations. TR$^2$ significantly outperforms previous state-of-the-art methods under two different settings by 2.1% and 2.6% respectively.

Abstract: Despite the remarkable success of convolutional neural networks in various computer vision tasks, recognizing indoor scenes still presents a significant challenge due to their complex composition. Consequently, effectively leveraging semantic information in the scene has been a key issue in advancing indoor scene recognition. Unfortunately, the accuracy of semantic segmentation has limited the effectiveness of existing approaches for leveraging semantic information. As a result, many of these approaches remain at the stage of auxiliary labeling or co-occurrence statistics, with few exploring the contextual relationships between the semantic elements directly within the scene. In this paper, we propose the Semantic Region Relationship Model (SRRM), which starts directly from the semantic information inside the scene. Specifically, SRRM adopts an adaptive and efficient approach to mitigate the negative impact of semantic ambiguity and then models the semantic region relationship to perform scene recognition. Additionally, to more comprehensively exploit the information contained in the scene, we combine the proposed SRRM with the PlacesCNN module to create the Combined Semantic Region Relation Model (CSRRM), and propose a novel information combining approach to effectively explore the complementary contents between them. CSRRM significantly outperforms the SOTA methods on the MIT Indoor 67, reduced Places365 dataset, and SUN RGB-D without retraining. The code is available at: https://github.com/ChuanxinSong/SRRM

Abstract: Deep convolutional neural networks have been pushing the frontier of face recognition (FR) techniques in the past years. Despite the high accuracy, they are often criticized for lacking explainability. There has been an increasing demand for understanding the decision-making process of deep face recognition systems. Recent studies have investigated using visual saliency maps as an explanation, but they often lack a discussion and analysis in the context of face recognition. This paper conceives a new explanation framework for face recognition. It starts by providing a new definition of the saliency-based explanation method, which focuses on the decisions made by the deep FR model. Then, a novel correlation-based RISE algorithm (CorrRISE) is proposed to produce saliency maps, which reveal both the similar and dissimilar regions of any given pair of face images. Besides, two evaluation metrics are designed to measure the performance of general visual saliency explanation methods in face recognition. Consequently, substantial visual and quantitative results have shown that the proposed method consistently outperforms other explainable face recognition approaches.

Abstract: Vision transformers (ViTs) achieve remarkable performance on large datasets, but tend to perform worse than convolutional neural networks (CNNs) when trained from scratch on smaller datasets, possibly due to a lack of local inductive bias in the architecture. Recent studies have therefore added locality to the architecture and demonstrated that it can help ViTs achieve performance comparable to CNNs in the small-size dataset regime. Existing methods, however, are architecture-specific or have higher computational and memory costs. Thus, we propose a module called Local InFormation Enhancer (LIFE) that extracts patch-level local information and incorporates it into the embeddings used in the self-attention block of ViTs. Our proposed module is memory and computation efficient, as well as flexible enough to process auxiliary tokens such as the classification and distillation tokens. Empirical results show that the addition of the LIFE module improves the performance of ViTs on small image classification datasets. We further demonstrate how the effect can be extended to downstream tasks, such as object detection and semantic segmentation. In addition, we introduce a new visualization method, Dense Attention Roll-Out, specifically designed for dense prediction tasks, allowing the generation of class-specific attention maps utilizing the attention maps of all tokens.

Abstract: Coordinate networks are widely used in computer vision due to their ability to represent signals as compressed, continuous entities. However, training these networks with first-order optimizers can be slow, hindering their use in real-time applications. Recent works have opted for shallow voxel-based representations to achieve faster training, but this sacrifices memory efficiency. This work proposes a solution that leverages second-order optimization methods to significantly reduce training times for coordinate networks while maintaining their compressibility. Experiments demonstrate the effectiveness of this approach on various signal modalities, such as audio, images, videos, shape reconstruction, and neural radiance fields.

Abstract: Long-term trajectory forecasting is a challenging problem in the field of computer vision and machine learning. In this paper, we propose a new method dubbed Di-Long ("Distillation for Long-Term trajectory") for long-term trajectory forecasting, which is based on knowledge distillation. Our approach involves training a student network to solve the long-term trajectory forecasting problem, whereas the teacher network from which the knowledge is distilled has a longer observation, and solves a short-term trajectory prediction problem by regularizing the student's predictions. Specifically, we use a teacher model to generate plausible trajectories for a shorter time horizon, and then distill the knowledge from the teacher model to a student model that solves the problem for a much higher time horizon. Our experiments show that the proposed Di-Long approach is beneficial for long-term forecasting, and our model achieves state-of-the-art performance on the Intersection Drone Dataset (inD) and the Stanford Drone Dataset (SDD).

Abstract: 3D convolutions are commonly employed by demosaicking neural models, in the same way as solving other image restoration problems. Counter-intuitively, we show that 3D convolutions implicitly impede the RGB color spectra from exchanging complementary information, resulting in spectral-inconsistent inference of the local spatial high frequency components. As a consequence, shallow 3D convolution networks suffer the Moir\'e artifacts, but deep 3D convolutions cause over-smoothness. We analyze the fundamental difference between demosaicking and other problems that predict lost pixels between available ones (e.g., super-resolution reconstruction), and present the underlying reasons for the confliction between Moir\'e-free and detail-preserving. From the new perspective, our work decouples the common standard convolution procedure to spectral and spatial feature aggregations, which allow strengthening global communication in the spectral dimension while respecting local contrast in the spatial dimension. We apply our demosaicking model to two tasks: Joint Demosaicking-Denoising and Independently Demosaicking. In both applications, our model substantially alleviates artifacts such as Moir\'e and over-smoothness at similar or lower computational cost to currently top-performing models, as validated by diverse evaluations. Source code will be released along with paper publication.

Abstract: With the progress of 3D human pose and shape estimation, state-of-the-art methods can either be robust to occlusions or obtain pixel-aligned accuracy in non-occlusion cases. However, they cannot obtain robustness and mesh-image alignment at the same time. In this work, we present NIKI (Neural Inverse Kinematics with Invertible Neural Network), which models bi-directional errors to improve the robustness to occlusions and obtain pixel-aligned accuracy. NIKI can learn from both the forward and inverse processes with invertible networks. In the inverse process, the model separates the error from the plausible 3D pose manifold for a robust 3D human pose estimation. In the forward process, we enforce the zero-error boundary conditions to improve the sensitivity to reliable joint positions for better mesh-image alignment. Furthermore, NIKI emulates the analytical inverse kinematics algorithms with the twist-and-swing decomposition for better interpretability. Experiments on standard and occlusion-specific benchmarks demonstrate the effectiveness of NIKI, where we exhibit robust and well-aligned results simultaneously. Code is available at https://github.com/Jeff-sjtu/NIKI

Abstract: Neural Architecture Search (NAS) aims to automatically excavate the optimal network architecture with superior test performance. Recent neural architecture search (NAS) approaches rely on validation loss or accuracy to find the superior network for the target data. In this paper, we investigate a new neural architecture search measure for excavating architectures with better generalization. We demonstrate that the flatness of the loss surface can be a promising proxy for predicting the generalization capability of neural network architectures. We evaluate our proposed method on various search spaces, showing similar or even better performance compared to the state-of-the-art NAS methods. Notably, the resultant architecture found by flatness measure generalizes robustly to various shifts in data distribution (e.g. ImageNet-V2,-A,-O), as well as various tasks such as object detection and semantic segmentation. Code is available at https://github.com/clovaai/GeNAS.

Abstract: Flows in networks (or graphs) play a significant role in numerous computer vision tasks. The scalar-valued edges in these graphs often lead to a loss of information and thereby to limitations in terms of expressiveness. For example, oftentimes high-dimensional data (e.g. feature descriptors) are mapped to a single scalar value (e.g. the similarity between two feature descriptors). To overcome this limitation, we propose a novel formalism for non-separable multi-dimensional network flows. By doing so, we enable an automatic and adaptive feature selection strategy - since the flow is defined on a per-dimension basis, the maximizing flow automatically chooses the best matching feature dimensions. As a proof of concept, we apply our formalism to the multi-object tracking problem and demonstrate that our approach outperforms scalar formulations on the MOT16 benchmark in terms of robustness to noise.

Abstract: In this paper, our goal is to design a simple learning paradigm for long-tail visual recognition, which not only improves the robustness of the feature extractor but also alleviates the bias of the classifier towards head classes while reducing the training skills and overhead. We propose an efficient one-stage training strategy for long-tailed visual recognition called Global and Local Mixture Consistency cumulative learning (GLMC). Our core ideas are twofold: (1) a global and local mixture consistency loss improves the robustness of the feature extractor. Specifically, we generate two augmented batches by the global MixUp and local CutMix from the same batch data, respectively, and then use cosine similarity to minimize the difference. (2) A cumulative head tail soft label reweighted loss mitigates the head class bias problem. We use empirical class frequencies to reweight the mixed label of the head-tail class for long-tailed data and then balance the conventional loss and the rebalanced loss with a coefficient accumulated by epochs. Our approach achieves state-of-the-art accuracy on CIFAR10-LT, CIFAR100-LT, and ImageNet-LT datasets. Additional experiments on balanced ImageNet and CIFAR demonstrate that GLMC can significantly improve the generalization of backbones. Code is made publicly available at https://github.com/ynu-yangpeng/GLMC.

Abstract: Following four successful years in the SAE AutoDrive Challenge Series I, the University of Toronto is participating in the Series II competition to develop a Level 4 autonomous passenger vehicle capable of handling various urban driving scenarios by 2025. Accurate detection of traffic lights and correct identification of their states is essential for safe autonomous operation in cities. Herein, we describe our recently-redesigned traffic light perception system for autonomous vehicles like the University of Toronto's self-driving car, Artemis. Similar to most traffic light perception systems, we rely primarily on camera-based object detectors. We deploy the YOLOv5 detector for bounding box regression and traffic light classification across multiple cameras and fuse the observations. To improve robustness, we incorporate priors from high-definition semantic maps and perform state filtering using hidden Markov models. We demonstrate a multi-camera, real time-capable traffic light perception pipeline that handles complex situations including multiple visible intersections, traffic light variations, temporary occlusion, and flashing light states. To validate our system, we collected and annotated a varied dataset incorporating flashing states and a range of occlusion types. Our results show superior performance in challenging real-world scenarios compared to single-frame, single-camera object detection.

Abstract: Contrastive learning has emerged as an efficient framework to learn multimodal representations. CLIP, a seminal work in this area, achieved impressive results by training on paired image-text data using the contrastive loss. Recent work claims improvements over CLIP using additional non-contrastive losses inspired from self-supervised learning. However, it is sometimes hard to disentangle the contribution of these additional losses from other implementation details, e.g., data augmentation or regularization techniques, used to train the model. To shed light on this matter, in this paper, we first propose, implement and evaluate several baselines obtained by combining contrastive learning with recent advances in self-supervised learning. In particular, we use the loss functions that were proven successful for visual self-supervised learning to align image and text modalities. We find that these baselines outperform a basic implementation of CLIP. However, when a stronger training recipe is employed, the advantage disappears. Indeed, we find that a simple CLIP baseline can also be improved substantially, up to a 25% relative improvement on downstream zero-shot tasks, by using well-known training techniques that are popular in other subfields. Moreover, we discover that it is enough to apply image and text augmentations to make up for most of the improvement attained by prior works. With our improved training recipe for CLIP, we obtain state-of-the-art performance on four standard datasets, and consistently outperform prior work (up to +4% on the largest dataset), while being substantially simpler.

Abstract: Visual Grounding (VG) refers to locating a region described by expressions in a specific image, which is a critical topic in vision-language fields. To alleviate the dependence on labeled data, existing unsupervised methods try to locate regions using task-unrelated pseudo-labels. However, a large proportion of pseudo-labels are noisy and diversity scarcity in language taxonomy. Inspired by the advances in V-L pretraining, we consider utilizing the VLP models to realize unsupervised transfer learning in downstream grounding task. Thus, we propose CLIP-VG, a novel method that can conduct self-paced curriculum adapting of CLIP via exploiting pseudo-language labels to solve VG problem. By elaborating an efficient model structure, we first propose a single-source and multi-source curriculum adapting method for unsupervised VG to progressively sample more reliable cross-modal pseudo-labels to obtain the optimal model, thus achieving implicit knowledge exploiting and denoising. Our method outperforms the existing state-of-the-art unsupervised VG method Pseudo-Q in both single-source and multi-source scenarios with a large margin, i.e., 6.78%~10.67% and 11.39%~24.87% on RefCOCO/+/g datasets, even outperforms existing weakly supervised methods. The code and models will be released at \url{https://github.com/linhuixiao/CLIP-VG}.

Abstract: Recently, Carlini et al. demonstrated the widely used model Stable Diffusion can regurgitate real training samples, which is troublesome from a copyright perspective. In this work, we provide an efficient extraction attack on par with the recent attack, with several order of magnitudes less network evaluations. In the process, we expose a new phenomena, which we dub template verbatims, wherein a diffusion model will regurgitate a training sample largely in tact. Template verbatims are harder to detect as they require retrieval and masking to correctly label. Furthermore, they are still generated by newer systems, even those which de-duplicate their training set, and we give insight into why they still appear during generation. We extract training images from several state of the art systems, including Stable Diffusion 2.0, Deep Image Floyd, and finally Midjourney v4. We release code to verify our extraction attack, perform the attack, as well as all extracted prompts at \url{https://github.com/ryanwebster90/onestep-extraction}.

Abstract: Document layout analysis is a crucial prerequisite for document understanding, including document retrieval and conversion. Most public datasets currently contain only PDF documents and lack realistic documents. Models trained on these datasets may not generalize well to real-world scenarios. Therefore, this paper introduces a large and diverse document layout analysis dataset called $M^{6}Doc$. The $M^6$ designation represents six properties: (1) Multi-Format (including scanned, photographed, and PDF documents); (2) Multi-Type (such as scientific articles, textbooks, books, test papers, magazines, newspapers, and notes); (3) Multi-Layout (rectangular, Manhattan, non-Manhattan, and multi-column Manhattan); (4) Multi-Language (Chinese and English); (5) Multi-Annotation Category (74 types of annotation labels with 237,116 annotation instances in 9,080 manually annotated pages); and (6) Modern documents. Additionally, we propose a transformer-based document layout analysis method called TransDLANet, which leverages an adaptive element matching mechanism that enables query embedding to better match ground truth to improve recall, and constructs a segmentation branch for more precise document image instance segmentation. We conduct a comprehensive evaluation of $M^{6}Doc$ with various layout analysis methods and demonstrate its effectiveness. TransDLANet achieves state-of-the-art performance on $M^{6}Doc$ with 64.5\% mAP. The $M^{6}Doc$ dataset will be available at https://github.com/HCIILAB/M6Doc.

Abstract: Few-shot learning for image classification comes up as a hot topic in computer vision, which aims at fast learning from a limited number of labeled images and generalize over the new tasks. In this paper, motivated by the idea of Fisher Score, we propose a Discriminative Local Descriptors Attention (DLDA) model that adaptively selects the representative local descriptors and does not introduce any additional parameters, while most of the existing local descriptors based methods utilize the neural networks that inevitably involve the tedious parameter tuning. Moreover, we modify the traditional $k$-NN classification model by adjusting the weights of the $k$ nearest neighbors according to their distances from the query point. Experiments on four benchmark datasets show that our method not only achieves higher accuracy compared with the state-of-art approaches for few-shot learning, but also possesses lower sensitivity to the choices of $k$.

Abstract: Foundation models have made significant strides in 2D and language tasks such as image segmentation, object detection, and visual-language understanding. Nevertheless, their potential to enhance 3D scene representation learning remains largely untapped due to the domain gap. In this paper, we propose an innovative methodology Bridge3D to address this gap, pre-training 3D models using features, semantic masks, and captions sourced from foundation models. Specifically, our approach utilizes semantic masks from these models to guide the masking and reconstruction process in the masked autoencoder. This strategy enables the network to concentrate more on foreground objects, thereby enhancing 3D representation learning. Additionally, we bridge the 3D-text gap at the scene level by harnessing image captioning foundation models. To further facilitate knowledge distillation from well-learned 2D and text representations to the 3D model, we introduce a novel method that employs foundation models to generate highly accurate object-level masks and semantic text information at the object level. Our approach notably outshines state-of-the-art methods in 3D object detection and semantic segmentation tasks. For instance, on the ScanNet dataset, our method surpasses the previous state-of-the-art method, PiMAE, by a significant margin of 5.3%.

Abstract: This paper proposes a novel age estimation algorithm, the Temporally-Aware Adaptive Graph Convolutional Network (TAA-GCN). Using a new representation based on graphs, the TAA-GCN utilizes skeletal, posture, clothing, and facial information to enrich the feature set associated with various ages. Such a novel graph representation has several advantages: First, reduced sensitivity to facial expression and other appearance variances; Second, robustness to partial occlusion and non-frontal-planar viewpoint, which is commonplace in real-world applications such as video surveillance. The TAA-GCN employs two novel components, (1) the Temporal Memory Module (TMM) to compute temporal dependencies in age; (2) Adaptive Graph Convolutional Layer (AGCL) to refine the graphs and accommodate the variance in appearance. The TAA-GCN outperforms the state-of-the-art methods on four public benchmarks, UTKFace, MORPHII, CACD, and FG-NET. Moreover, the TAA-GCN showed reliability in different camera viewpoints and reduced quality images.

Abstract: This paper tries to address a fundamental question in point cloud self-supervised learning: what is a good signal we should leverage to learn features from point clouds without annotations? To answer that, we introduce a point cloud representation learning framework, based on geometric feature reconstruction. In contrast to recent papers that directly adopt masked autoencoder (MAE) and only predict original coordinates or occupancy from masked point clouds, our method revisits differences between images and point clouds and identifies three self-supervised learning objectives peculiar to point clouds, namely centroid prediction, normal estimation, and curvature prediction. Combined with occupancy prediction, these four objectives yield an nontrivial self-supervised learning task and mutually facilitate models to better reason fine-grained geometry of point clouds. Our pipeline is conceptually simple and it consists of two major steps: first, it randomly masks out groups of points, followed by a Transformer-based point cloud encoder; second, a lightweight Transformer decoder predicts centroid, normal, and curvature for points in each voxel. We transfer the pre-trained Transformer encoder to a downstream peception model. On the nuScene Datset, our model achieves 3.38 mAP improvment for object detection, 2.1 mIoU gain for segmentation, and 1.7 AMOTA gain for multi-object tracking. We also conduct experiments on the Waymo Open Dataset and achieve significant performance improvements over baselines as well.

Abstract: A fully automated object reconstruction pipeline is crucial for digital content creation. While the area of 3D reconstruction has witnessed profound developments, the removal of background to obtain a clean object model still relies on different forms of manual labor, such as bounding box labeling, mask annotations, and mesh manipulations. In this paper, we propose a novel framework named AutoRecon for the automated discovery and reconstruction of an object from multi-view images. We demonstrate that foreground objects can be robustly located and segmented from SfM point clouds by leveraging self-supervised 2D vision transformer features. Then, we reconstruct decomposed neural scene representations with dense supervision provided by the decomposed point clouds, resulting in accurate object reconstruction and segmentation. Experiments on the DTU, BlendedMVS and CO3D-V2 datasets demonstrate the effectiveness and robustness of AutoRecon.

Abstract: A lightweight underwater image enhancement network is of great significance for resource-constrained platforms, but balancing model size, computational efficiency, and enhancement performance has proven difficult for previous approaches. In this work, we propose the Five A$^{+}$ Network (FA$^{+}$Net), a highly efficient and lightweight real-time underwater image enhancement network with only $\sim$ 9k parameters and $\sim$ 0.01s processing time. The FA$^{+}$Net employs a two-stage enhancement structure. The strong prior stage aims to decompose challenging underwater degradations into sub-problems, while the fine-grained stage incorporates multi-branch color enhancement module and pixel attention module to amplify the network's perception of details. To the best of our knowledge, FA$^{+}$Net is the only network with the capability of real-time enhancement of 1080P images. Thorough extensive experiments and comprehensive visual comparison, we show that FA$^{+}$Net outperforms previous approaches by obtaining state-of-the-art performance on multiple datasets while significantly reducing both parameter count and computational complexity. The code is open source at https://github.com/Owen718/FiveAPlus-Network.

Abstract: Recent self-supervised contrastive learning methods greatly benefit from the Siamese structure that aims to minimizing distances between positive pairs. These methods usually apply random data augmentation to input images, expecting the augmented views of the same images to be similar and positively paired. However, random augmentation may overlook image semantic information and degrade the quality of augmented views in contrastive learning. This issue becomes more challenging in medical images since the abnormalities related to diseases can be tiny, and are easy to be corrupted (e.g., being cropped out) in the current scheme of random augmentation. In this work, we first demonstrate that, for widely-used X-ray images, the conventional augmentation prevalent in contrastive pre-training can affect the performance of the downstream diagnosis or classification tasks. Then, we propose a novel augmentation method, i.e., FocusContrast, to learn from radiologists' gaze in diagnosis and generate contrastive views for medical images with guidance from radiologists' visual attention. Specifically, we track the gaze movement of radiologists and model their visual attention when reading to diagnose X-ray images. The learned model can predict visual attention of the radiologists given a new input image, and further guide the attention-aware augmentation that hardly neglects the disease-related abnormalities. As a plug-and-play and framework-agnostic module, FocusContrast consistently improves state-of-the-art contrastive learning methods of SimCLR, MoCo, and BYOL by 4.0~7.0% in classification accuracy on a knee X-ray dataset.

Abstract: Perceptual similarity metrics have progressively become more correlated with human judgments on perceptual similarity; however, despite recent advances, the addition of an imperceptible distortion can still compromise these metrics. In our study, we systematically examine the robustness of these metrics to imperceptible adversarial perturbations. Following the two-alternative forced-choice experimental design with two distorted images and one reference image, we perturb the distorted image closer to the reference via an adversarial attack until the metric flips its judgment. We first show that all metrics in our study are susceptible to perturbations generated via common adversarial attacks such as FGSM, PGD, and the One-pixel attack. Next, we attack the widely adopted LPIPS metric using spatial-transformation-based adversarial perturbations (stAdv) in a white-box setting to craft adversarial examples that can effectively transfer to other similarity metrics in a black-box setting. We also combine the spatial attack stAdv with PGD ($\ell_\infty$-bounded) attack to increase transferability and use these adversarial examples to benchmark the robustness of both traditional and recently developed metrics. Our benchmark provides a good starting point for discussion and further research on the robustness of metrics to imperceptible adversarial perturbations.

Abstract: The text-driven image and video diffusion models have achieved unprecedented success in generating realistic and diverse content. Recently, the editing and variation of existing images and videos in diffusion-based generative models have garnered significant attention. However, previous works are limited to editing content with text or providing coarse personalization using a single visual clue, rendering them unsuitable for indescribable content that requires fine-grained and detailed control. In this regard, we propose a generic video editing framework called Make-A-Protagonist, which utilizes textual and visual clues to edit videos with the goal of empowering individuals to become the protagonists. Specifically, we leverage multiple experts to parse source video, target visual and textual clues, and propose a visual-textual-based video generation model that employs mask-guided denoising sampling to generate the desired output. Extensive results demonstrate the versatile and remarkable editing capabilities of Make-A-Protagonist.

Abstract: While bird's-eye-view (BEV) perception models can be useful for building high-definition maps (HD-Maps) with less human labor, their results are often unreliable and demonstrate noticeable inconsistencies in the predicted HD-Maps from different viewpoints. This is because BEV perception is typically set up in an 'onboard' manner, which restricts the computation and consequently prevents algorithms from reasoning multiple views simultaneously. This paper overcomes these limitations and advocates a more practical 'offboard' HD-Map generation setup that removes the computation constraints, based on the fact that HD-Maps are commonly reusable infrastructures built offline in data centers. To this end, we propose a novel offboard pipeline called MV-Map that capitalizes multi-view consistency and can handle an arbitrary number of frames with the key design of a 'region-centric' framework. In MV-Map, the target HD-Maps are created by aggregating all the frames of onboard predictions, weighted by the confidence scores assigned by an 'uncertainty network'. To further enhance multi-view consistency, we augment the uncertainty network with the global 3D structure optimized by a voxelized neural radiance field (Voxel-NeRF). Extensive experiments on nuScenes show that our MV-Map significantly improves the quality of HD-Maps, further highlighting the importance of offboard methods for HD-Map generation.

Abstract: Speech-driven animation has gained significant traction in recent years, with current methods achieving near-photorealistic results. However, the field remains underexplored regarding non-verbal communication despite evidence demonstrating its importance in human interaction. In particular, generating laughter sequences presents a unique challenge due to the intricacy and nuances of this behaviour. This paper aims to bridge this gap by proposing a novel model capable of generating realistic laughter sequences, given a still portrait and an audio clip containing laughter. We highlight the failure cases of traditional facial animation methods and leverage recent advances in diffusion models to produce convincing laughter videos. We train our model on a diverse set of laughter datasets and introduce an evaluation metric specifically designed for laughter. When compared with previous speech-driven approaches, our model achieves state-of-the-art performance across all metrics, even when these are re-trained for laughter generation.