Computer Vision and Pattern Recognition (cs.CV)

Abstract: Story visualization (SV) is a challenging text-to-image generation task for the difficulty of not only rendering visual details from the text descriptions but also encoding a long-term context across multiple sentences. While prior efforts mostly focus on generating a semantically relevant image for each sentence, encoding a context spread across the given paragraph to generate contextually convincing images (e.g., with a correct character or with a proper background of the scene) remains a challenge. To this end, we propose a novel memory architecture for the Bi-directional Transformers with an online text augmentation that generates multiple pseudo-descriptions as supplementary supervision during training, for better generalization to the language variation at inference. In extensive experiments on the two popular SV benchmarks, i.e., the Pororo-SV and Flintstones-SV, the proposed method significantly outperforms the state of the arts in various evaluation metrics including FID, character F1, frame accuracy, BLEU-2/3, and R-precision with similar or less computational complexity.

Abstract: Cycling stress assessment, which quantifies cyclists' perceived stress imposed by the built environment and motor traffics, increasingly informs cycling infrastructure planning and cycling route recommendation. However, currently calculating cycling stress is slow and data-intensive, which hinders its broader application. In this paper, We propose a deep learning framework to support accurate, fast, and large-scale cycling stress assessments for urban road networks based on street-view images. Our framework features i) a contrastive learning approach that leverages the ordinal relationship among cycling stress labels, and ii) a post-processing technique that enforces spatial smoothness into our predictions. On a dataset of 39,153 road segments collected in Toronto, Canada, our results demonstrate the effectiveness of our deep learning framework and the value of using image data for cycling stress assessment in the absence of high-quality road geometry and motor traffic data.

Abstract: Autonomous driving systems require many images for analyzing the surrounding environment. However, there is fewer data protection for private information among these captured images, such as pedestrian faces or vehicle license plates, which has become a significant issue. In this paper, in response to the call for data security laws and regulations and based on the advantages of large Field of View(FoV) of the fisheye camera, we build the first Autopilot Desensitization Dataset, called ADD, and formulate the first deep-learning-based image desensitization framework, to promote the study of image desensitization in autonomous driving scenarios. The compiled dataset consists of 650K images, including different face and vehicle license plate information captured by the surround-view fisheye camera. It covers various autonomous driving scenarios, including diverse facial characteristics and license plate colors. Then, we propose an efficient multitask desensitization network called DesCenterNet as a benchmark on the ADD dataset, which can perform face and vehicle license plate detection and desensitization tasks. Based on ADD, we further provide an evaluation criterion for desensitization performance, and extensive comparison experiments have verified the effectiveness and superiority of our method on image desensitization.

Abstract: The transformer-based semantic segmentation approaches, which divide the image into different regions by sliding windows and model the relation inside each window, have achieved outstanding success. However, since the relation modeling between windows was not the primary emphasis of previous work, it was not fully utilized. To address this issue, we propose a Graph-Segmenter, including a Graph Transformer and a Boundary-aware Attention module, which is an effective network for simultaneously modeling the more profound relation between windows in a global view and various pixels inside each window as a local one, and for substantial low-cost boundary adjustment. Specifically, we treat every window and pixel inside the window as nodes to construct graphs for both views and devise the Graph Transformer. The introduced boundary-aware attention module optimizes the edge information of the target objects by modeling the relationship between the pixel on the object's edge. Extensive experiments on three widely used semantic segmentation datasets (Cityscapes, ADE-20k and PASCAL Context) demonstrate that our proposed network, a Graph Transformer with Boundary-aware Attention, can achieve state-of-the-art segmentation performance.

Abstract: Visual Speech Recognition (VSR) is the task of predicting spoken words from silent lip movements. VSR is regarded as a challenging task because of the insufficient information on lip movements. In this paper, we propose an Audio Knowledge empowered Visual Speech Recognition framework (AKVSR) to complement the insufficient speech information of visual modality by using audio modality. Different from the previous methods, the proposed AKVSR 1) utilizes rich audio knowledge encoded by a large-scale pretrained audio model, 2) saves the linguistic information of audio knowledge in compact audio memory by discarding the non-linguistic information from the audio through quantization, and 3) includes Audio Bridging Module which can find the best-matched audio features from the compact audio memory, which makes our training possible without audio inputs, once after the compact audio memory is composed. We validate the effectiveness of the proposed method through extensive experiments, and achieve new state-of-the-art performances on the widely-used datasets, LRS2 and LRS3.

Abstract: This paper reports on the development of \textbf{a novel style guided diffusion model (SGDiff)} which overcomes certain weaknesses inherent in existing models for image synthesis. The proposed SGDiff combines image modality with a pretrained text-to-image diffusion model to facilitate creative fashion image synthesis. It addresses the limitations of text-to-image diffusion models by incorporating supplementary style guidance, substantially reducing training costs, and overcoming the difficulties of controlling synthesized styles with text-only inputs. This paper also introduces a new dataset -- SG-Fashion, specifically designed for fashion image synthesis applications, offering high-resolution images and an extensive range of garment categories. By means of comprehensive ablation study, we examine the application of classifier-free guidance to a variety of conditions and validate the effectiveness of the proposed model for generating fashion images of the desired categories, product attributes, and styles. The contributions of this paper include a novel classifier-free guidance method for multi-modal feature fusion, a comprehensive dataset for fashion image synthesis application, a thorough investigation on conditioned text-to-image synthesis, and valuable insights for future research in the text-to-image synthesis domain. The code and dataset are available at: \url{https://github.com/taited/SGDiff}.

Abstract: We present a method for learning multiple scene representations given a small labeled set, by exploiting the relationships between such representations in the form of a multi-task hypergraph. We also show how we can use the hypergraph to improve a powerful pretrained VisTransformer model without any additional labeled data. In our hypergraph, each node is an interpretation layer (e.g., depth or segmentation) of the scene. Within each hyperedge, one or several input nodes predict the layer at the output node. Thus, each node could be an input node in some hyperedges and an output node in others. In this way, multiple paths can reach the same node, to form ensembles from which we obtain robust pseudolabels, which allow self-supervised learning in the hypergraph. We test different ensemble models and different types of hyperedges and show superior performance to other multi-task graph models in the field. We also introduce Dronescapes, a large video dataset captured with UAVs in different complex real-world scenes, with multiple representations, suitable for multi-task learning.

Abstract: Recent advancements in large foundation models have shown promising potential in the medical industry due to their flexible prompting capability. One such model, the Segment Anything Model (SAM), a prompt-driven segmentation model, has shown remarkable performance improvements, surpassing state-of-the-art approaches in medical image segmentation. However, existing methods primarily rely on tuning strategies that require extensive data or prior prompts tailored to the specific task, making it particularly challenging when only a limited number of data samples are available. In this paper, we propose a novel perspective on self-prompting in medical vision applications. Specifically, we harness the embedding space of SAM to prompt itself through a simple yet effective linear pixel-wise classifier. By preserving the encoding capabilities of the large model, the contextual information from its decoder, and leveraging its interactive promptability, we achieve competitive results on multiple datasets (i.e. improvement of more than 15% compared to fine-tuning the mask decoder using a few images).

Abstract: Deep neural networks are vulnerable to adversarial examples, dictating the imperativeness to test the model's robustness before deployment. Transfer-based attackers craft adversarial examples against surrogate models and transfer them to victim models deployed in the black-box situation. To enhance the adversarial transferability, structure-based attackers adjust the backpropagation path to avoid the attack from overfitting the surrogate model. However, existing structure-based attackers fail to explore the convolution module in CNNs and modify the backpropagation graph heuristically, leading to limited effectiveness. In this paper, we propose backPropagation pAth Search (PAS), solving the aforementioned two problems. We first propose SkipConv to adjust the backpropagation path of convolution by structural reparameterization. To overcome the drawback of heuristically designed backpropagation paths, we further construct a DAG-based search space, utilize one-step approximation for path evaluation and employ Bayesian Optimization to search for the optimal path. We conduct comprehensive experiments in a wide range of transfer settings, showing that PAS improves the attack success rate by a huge margin for both normally trained and defense models.

Abstract: In text-video retrieval, recent works have benefited from the powerful learning capabilities of pre-trained text-image foundation models (e.g., CLIP) by adapting them to the video domain. A critical problem for them is how to effectively capture the rich semantics inside the video using the image encoder of CLIP. To tackle this, state-of-the-art methods adopt complex cross-modal modeling techniques to fuse the text information into video frame representations, which, however, incurs severe efficiency issues in large-scale retrieval systems as the video representations must be recomputed online for every text query. In this paper, we discard this problematic cross-modal fusion process and aim to learn semantically-enhanced representations purely from the video, so that the video representations can be computed offline and reused for different texts. Concretely, we first introduce a spatial-temporal "Prompt Cube" into the CLIP image encoder and iteratively switch it within the encoder layers to efficiently incorporate the global video semantics into frame representations. We then propose to apply an auxiliary video captioning objective to train the frame representations, which facilitates the learning of detailed video semantics by providing fine-grained guidance in the semantic space. With a naive temporal fusion strategy (i.e., mean-pooling) on the enhanced frame representations, we obtain state-of-the-art performances on three benchmark datasets, i.e., MSR-VTT, MSVD, and LSMDC.

Abstract: Current model quantization methods have shown their promising capability in reducing storage space and computation complexity. However, due to the diversity of quantization forms supported by different hardware, one limitation of existing solutions is that usually require repeated optimization for different scenarios. How to construct a model with flexible quantization forms has been less studied. In this paper, we explore a one-shot network quantization regime, named Elastic Quantization Neural Networks (EQ-Net), which aims to train a robust weight-sharing quantization supernet. First of all, we propose an elastic quantization space (including elastic bit-width, granularity, and symmetry) to adapt to various mainstream quantitative forms. Secondly, we propose the Weight Distribution Regularization Loss (WDR-Loss) and Group Progressive Guidance Loss (GPG-Loss) to bridge the inconsistency of the distribution for weights and output logits in the elastic quantization space gap. Lastly, we incorporate genetic algorithms and the proposed Conditional Quantization-Aware Accuracy Predictor (CQAP) as an estimator to quickly search mixed-precision quantized neural networks in supernet. Extensive experiments demonstrate that our EQ-Net is close to or even better than its static counterparts as well as state-of-the-art robust bit-width methods. Code can be available at \href{https://github.com/xuke225/EQ-Net.git}{https://github.com/xuke225/EQ-Net}.

Abstract: Equipping the rototranslation group $SE(2)$ with a sub-Riemannian structure inspired by the visual cortex V1, we propose algorithms for image inpainting and enhancement based on hypoelliptic diffusion. We innovate on previous implementations of the methods by Citti, Sarti and Boscain et al., by proposing an alternative that prevents fading and capable of producing sharper results in a procedure that we call WaxOn-WaxOff. We also exploit the sub-Riemannian structure to define a completely new unsharp using $SE(2)$, analogous of the classical unsharp filter for 2D image processing, with applications to image enhancement. We demonstrate our method on blood vessels enhancement in retinal scans.

Abstract: Exemplar-based sketch-to-photo synthesis allows users to generate photo-realistic images based on sketches. Recently, diffusion-based methods have achieved impressive performance on image generation tasks, enabling highly-flexible control through text-driven generation or energy functions. However, generating photo-realistic images with color and texture from sketch images remains challenging for diffusion models. Sketches typically consist of only a few strokes, with most regions left blank, making it difficult for diffusion-based methods to produce photo-realistic images. In this work, we propose a two-stage method named ``Inversion-by-Inversion" for exemplar-based sketch-to-photo synthesis. This approach includes shape-enhancing inversion and full-control inversion. During the shape-enhancing inversion process, an uncolored photo is generated with the guidance of a shape-energy function. This step is essential to ensure control over the shape of the generated photo. In the full-control inversion process, we propose an appearance-energy function to control the color and texture of the final generated photo.Importantly, our Inversion-by-Inversion pipeline is training-free and can accept different types of exemplars for color and texture control. We conducted extensive experiments to evaluate our proposed method, and the results demonstrate its effectiveness.

Abstract: Deep neural networks have been widely used in various downstream tasks, especially those safety-critical scenario such as autonomous driving, but deep networks are often threatened by adversarial samples. Such adversarial attacks can be invisible to human eyes, but can lead to DNN misclassification, and often exhibits transferability between deep learning and machine learning models and real-world achievability. Adversarial attacks can be divided into white-box attacks, for which the attacker knows the parameters and gradient of the model, and black-box attacks, for the latter, the attacker can only obtain the input and output of the model. In terms of the attacker's purpose, it can be divided into targeted attacks and non-targeted attacks, which means that the attacker wants the model to misclassify the original sample into the specified class, which is more practical, while the non-targeted attack just needs to make the model misclassify the sample. The black box setting is a scenario we will encounter in practice.

Abstract: While the field of multi-modal learning keeps growing fast, the deficiency of the standard joint training paradigm has become clear through recent studies. They attribute the sub-optimal performance of the jointly trained model to the modality competition phenomenon. Existing works attempt to improve the jointly trained model by modulating the training process. Despite their effectiveness, those methods can only apply to late fusion models. More importantly, the mechanism of the modality competition remains unexplored. In this paper, we first propose an adaptive gradient modulation method that can boost the performance of multi-modal models with various fusion strategies. Extensive experiments show that our method surpasses all existing modulation methods. Furthermore, to have a quantitative understanding of the modality competition and the mechanism behind the effectiveness of our modulation method, we introduce a novel metric to measure the competition strength. This metric is built on the mono-modal concept, a function that is designed to represent the competition-less state of a modality. Through systematic investigation, our results confirm the intuition that the modulation encourages the model to rely on the more informative modality. In addition, we find that the jointly trained model typically has a preferred modality on which the competition is weaker than other modalities. However, this preferred modality need not dominate others. Our code will be available at https://github.com/lihong2303/AGM_ICCV2023.

Abstract: Given a classifier, the inherent property of semantic Out-of-Distribution (OOD) samples is that their contents differ from all legal classes in terms of semantics, namely semantic mismatch. There is a recent work that directly applies it to OOD detection, which employs a conditional Generative Adversarial Network (cGAN) to enlarge semantic mismatch in the image space. While achieving remarkable OOD detection performance on small datasets, it is not applicable to ImageNet-scale datasets due to the difficulty in training cGANs with both input images and labels as conditions. As diffusion models are much easier to train and amenable to various conditions compared to cGANs, in this work, we propose to directly use pre-trained diffusion models for semantic mismatch-guided OOD detection, named DiffGuard. Specifically, given an OOD input image and the predicted label from the classifier, we try to enlarge the semantic difference between the reconstructed OOD image under these conditions and the original input image. We also present several test-time techniques to further strengthen such differences. Experimental results show that DiffGuard is effective on both Cifar-10 and hard cases of the large-scale ImageNet, and it can be easily combined with existing OOD detection techniques to achieve state-of-the-art OOD detection results.

Abstract: Medical Image Segmentation is crucial in various clinical applications within the medical domain. While state-of-the-art segmentation models have proven effective, integrating textual guidance to enhance visual features for this task remains an area with limited progress. Existing segmentation models that utilize textual guidance are primarily trained on open-domain images, raising concerns about their direct applicability in the medical domain without manual intervention or fine-tuning. To address these challenges, we propose using multimodal vision-language models for capturing semantic information from image descriptions and images, enabling the segmentation of diverse medical images. This study comprehensively evaluates existing vision language models across multiple datasets to assess their transferability from the open domain to the medical field. Furthermore, we introduce variations of image descriptions for previously unseen images in the dataset, revealing notable variations in model performance based on the generated prompts. Our findings highlight the distribution shift between the open-domain images and the medical domain and show that the segmentation models trained on open-domain images are not directly transferrable to the medical field. But their performance can be increased by finetuning them in the medical datasets. We report the zero-shot and finetuned segmentation performance of 4 Vision Language Models (VLMs) on 11 medical datasets using 9 types of prompts derived from 14 attributes.

Abstract: Motion mode (M-mode) recording is an essential part of echocardiography to measure cardiac dimension and function. However, the current diagnosis cannot build an automatic scheme, as there are three fundamental obstructs: Firstly, there is no open dataset available to build the automation for ensuring constant results and bridging M-mode echocardiography with real-time instance segmentation (RIS); Secondly, the examination is involving the time-consuming manual labelling upon M-mode echocardiograms; Thirdly, as objects in echocardiograms occupy a significant portion of pixels, the limited receptive field in existing backbones (e.g., ResNet) composed from multiple convolution layers are inefficient to cover the period of a valve movement. Existing non-local attentions (NL) compromise being unable real-time with a high computation overhead or losing information from a simplified version of the non-local block. Therefore, we proposed RAMEM, a real-time automatic M-mode echocardiography measurement scheme, contributes three aspects to answer the problems: 1) provide MEIS, a dataset of M-mode echocardiograms for instance segmentation, to enable consistent results and support the development of an automatic scheme; 2) propose panel attention, local-to-global efficient attention by pixel-unshuffling, embedding with updated UPANets V2 in a RIS scheme toward big object detection with global receptive field; 3) develop and implement AMEM, an efficient algorithm of automatic M-mode echocardiography measurement enabling fast and accurate automatic labelling among diagnosis. The experimental results show that RAMEM surpasses existing RIS backbones (with non-local attention) in PASCAL 2012 SBD and human performances in real-time MEIS tested. The code of MEIS and dataset are available at https://github.com/hanktseng131415go/RAME.

Abstract: In the domain adaptation problem, source data may be unavailable to the target client side due to privacy or intellectual property issues. Source-free unsupervised domain adaptation (SF-UDA) aims at adapting a model trained on the source side to align the target distribution with only the source model and unlabeled target data. The source model usually produces noisy and context-inconsistent pseudo-labels on the target domain, i.e., neighbouring regions that have a similar visual appearance are annotated with different pseudo-labels. This observation motivates us to refine pseudo-labels with context relations. Another observation is that features of the same class tend to form a cluster despite the domain gap, which implies context relations can be readily calculated from feature distances. To this end, we propose a context-aware pseudo-label refinement method for SF-UDA. Specifically, a context-similarity learning module is developed to learn context relations. Next, pseudo-label revision is designed utilizing the learned context relations. Further, we propose calibrating the revised pseudo-labels to compensate for wrong revision caused by inaccurate context relations. Additionally, we adopt a pixel-level and class-level denoising scheme to select reliable pseudo-labels for domain adaptation. Experiments on cross-domain fundus images indicate that our approach yields the state-of-the-art results. Code is available at https://github.com/xmed-lab/CPR.

Abstract: Jointly processing information from multiple sensors is crucial to achieving accurate and robust perception for reliable autonomous driving systems. However, current 3D perception research follows a modality-specific paradigm, leading to additional computation overheads and inefficient collaboration between different sensor data. In this paper, we present an efficient multi-modal backbone for outdoor 3D perception named UniTR, which processes a variety of modalities with unified modeling and shared parameters. Unlike previous works, UniTR introduces a modality-agnostic transformer encoder to handle these view-discrepant sensor data for parallel modal-wise representation learning and automatic cross-modal interaction without additional fusion steps. More importantly, to make full use of these complementary sensor types, we present a novel multi-modal integration strategy by both considering semantic-abundant 2D perspective and geometry-aware 3D sparse neighborhood relations. UniTR is also a fundamentally task-agnostic backbone that naturally supports different 3D perception tasks. It sets a new state-of-the-art performance on the nuScenes benchmark, achieving +1.1 NDS higher for 3D object detection and +12.0 higher mIoU for BEV map segmentation with lower inference latency. Code will be available at https://github.com/Haiyang-W/UniTR .

Abstract: In Video Object Detection (VID), a common practice is to leverage the rich temporal contexts from the video to enhance the object representations in each frame. Existing methods treat the temporal contexts obtained from different objects indiscriminately and ignore their different identities. While intuitively, aggregating local views of the same object in different frames may facilitate a better understanding of the object. Thus, in this paper, we aim to enable the model to focus on the identity-consistent temporal contexts of each object to obtain more comprehensive object representations and handle the rapid object appearance variations such as occlusion, motion blur, etc. However, realizing this goal on top of existing VID models faces low-efficiency problems due to their redundant region proposals and nonparallel frame-wise prediction manner. To aid this, we propose ClipVID, a VID model equipped with Identity-Consistent Aggregation (ICA) layers specifically designed for mining fine-grained and identity-consistent temporal contexts. It effectively reduces the redundancies through the set prediction strategy, making the ICA layers very efficient and further allowing us to design an architecture that makes parallel clip-wise predictions for the whole video clip. Extensive experimental results demonstrate the superiority of our method: a state-of-the-art (SOTA) performance (84.7% mAP) on the ImageNet VID dataset while running at a speed about 7x faster (39.3 fps) than previous SOTAs.

Abstract: Visual chart recognition systems are gaining increasing attention due to the growing demand for automatically identifying table headers and values from chart images. Current methods rely on keypoint detection to estimate data element shapes in charts but suffer from grouping errors in post-processing. To address this issue, we propose ChartDETR, a transformer-based multi-shape detector that localizes keypoints at the corners of regular shapes to reconstruct multiple data elements in a single chart image. Our method predicts all data element shapes at once by introducing query groups in set prediction, eliminating the need for further postprocessing. This property allows ChartDETR to serve as a unified framework capable of representing various chart types without altering the network architecture, effectively detecting data elements of diverse shapes. We evaluated ChartDETR on three datasets, achieving competitive results across all chart types without any additional enhancements. For example, ChartDETR achieved an F1 score of 0.98 on Adobe Synthetic, significantly outperforming the previous best model with a 0.71 F1 score. Additionally, we obtained a new state-of-the-art result of 0.97 on ExcelChart400k. The code will be made publicly available.

Abstract: Having precise perception of the environment is crucial for ensuring the secure and reliable functioning of autonomous driving systems. Radar object detection networks are one fundamental part of such systems. CNN-based object detectors showed good performance in this context, but they require large compute resources. This paper investigates sparse convolutional object detection networks, which combine powerful grid-based detection with low compute resources. We investigate radar specific challenges and propose sparse kernel point pillars (SKPP) and dual voxel point convolutions (DVPC) as remedies for the grid rendering and sparse backbone architectures. We evaluate our SKPP-DPVCN architecture on nuScenes, which outperforms the baseline by 5.89% and the previous state of the art by 4.19% in Car AP4.0. Moreover, SKPP-DPVCN reduces the average scale error (ASE) by 21.41% over the baseline.

Abstract: The rising demand for creating lifelike avatars in the digital realm has led to an increased need for generating high-quality human videos guided by textual descriptions and poses. We propose Dancing Avatar, designed to fabricate human motion videos driven by poses and textual cues. Our approach employs a pretrained T2I diffusion model to generate each video frame in an autoregressive fashion. The crux of innovation lies in our adept utilization of the T2I diffusion model for producing video frames successively while preserving contextual relevance. We surmount the hurdles posed by maintaining human character and clothing consistency across varying poses, along with upholding the background's continuity amidst diverse human movements. To ensure consistent human appearances across the entire video, we devise an intra-frame alignment module. This module assimilates text-guided synthesized human character knowledge into the pretrained T2I diffusion model, synergizing insights from ChatGPT. For preserving background continuity, we put forth a background alignment pipeline, amalgamating insights from segment anything and image inpainting techniques. Furthermore, we propose an inter-frame alignment module that draws inspiration from an auto-regressive pipeline to augment temporal consistency between adjacent frames, where the preceding frame guides the synthesis process of the current frame. Comparisons with state-of-the-art methods demonstrate that Dancing Avatar exhibits the capacity to generate human videos with markedly superior quality, both in terms of human and background fidelity, as well as temporal coherence compared to existing state-of-the-art approaches.

Abstract: The prediction of road users' future motion is a critical task in supporting advanced driver-assistance systems (ADAS). It plays an even more crucial role for autonomous driving (AD) in enabling the planning and execution of safe driving maneuvers. Based on our previous work, Context-Aware Scene Prediction Network (CASPNet), an improved system, CASPNet++, is proposed. In this work, we focus on further enhancing the interaction modeling and scene understanding to support the joint prediction of all road users in a scene using spatiotemporal grids to model future occupancy. Moreover, an instance-based output head is introduced to provide multi-modal trajectories for agents of interest. In extensive quantitative and qualitative analysis, we demonstrate the scalability of CASPNet++ in utilizing and fusing diverse environmental input sources such as HD maps, Radar detection, and Lidar segmentation. Tested on the urban-focused prediction dataset nuScenes, CASPNet++ reaches state-of-the-art performance. The model has been deployed in a testing vehicle, running in real-time with moderate computational resources.

Abstract: With a number of marine populations in rapid decline, collecting and analyzing data about marine populations has become increasingly important to develop effective conservation policies for a wide range of marine animals, including whales. Modern computer vision algorithms allow us to detect whales in images in a wide range of domains, further speeding up and enhancing the monitoring process. However, these algorithms heavily rely on large training datasets, which are challenging and time-consuming to collect particularly in marine or aquatic environments. Recent advances in AI however have made it possible to synthetically create datasets for training machine learning algorithms, thus enabling new solutions that were not possible before. In this work, we present a solution - SeaDroneSim2 benchmark suite, which addresses this challenge by generating aerial, and satellite synthetic image datasets to improve the detection of whales and reduce the effort required for training data collection. We show that we can achieve a 15% performance boost on whale detection compared to using the real data alone for training, by augmenting a 10% real data. We open source both the code of the simulation platform SeaDroneSim2 and the dataset generated through it.

Abstract: Facial Action Unit (AU) detection is a crucial task in affective computing and social robotics as it helps to identify emotions expressed through facial expressions. Anatomically, there are innumerable correlations between AUs, which contain rich information and are vital for AU detection. Previous methods used fixed AU correlations based on expert experience or statistical rules on specific benchmarks, but it is challenging to comprehensively reflect complex correlations between AUs via hand-crafted settings. There are alternative methods that employ a fully connected graph to learn these dependencies exhaustively. However, these approaches can result in a computational explosion and high dependency with a large dataset. To address these challenges, this paper proposes a novel self-adjusting AU-correlation learning (SACL) method with less computation for AU detection. This method adaptively learns and updates AU correlation graphs by efficiently leveraging the characteristics of different levels of AU motion and emotion representation information extracted in different stages of the network. Moreover, this paper explores the role of multi-scale learning in correlation information extraction, and design a simple yet effective multi-scale feature learning (MSFL) method to promote better performance in AU detection. By integrating AU correlation information with multi-scale features, the proposed method obtains a more robust feature representation for the final AU detection. Extensive experiments show that the proposed method outperforms the state-of-the-art methods on widely used AU detection benchmark datasets, with only 28.7\% and 12.0\% of the parameters and FLOPs of the best method, respectively. The code for this method is available at \url{https://github.com/linuxsino/Self-adjusting-AU}.

Abstract: Remote photoplethysmography (rPPG) based physiological measurement is an emerging yet crucial vision task, whose challenge lies in exploring accurate rPPG prediction from facial videos accompanied by noises of illumination variations, facial occlusions, head movements, \etc, in a non-contact manner. Existing mainstream CNN-based models make efforts to detect physiological signals by capturing subtle color changes in facial regions of interest (ROI) caused by heartbeats. However, such models are constrained by the limited local spatial or temporal receptive fields in the neural units. Unlike them, a native Transformer-based framework called Dual-path TokenLearner (Dual-TL) is proposed in this paper, which utilizes the concept of learnable tokens to integrate both spatial and temporal informative contexts from the global perspective of the video. Specifically, the proposed Dual-TL uses a Spatial TokenLearner (S-TL) to explore associations in different facial ROIs, which promises the rPPG prediction far away from noisy ROI disturbances. Complementarily, a Temporal TokenLearner (T-TL) is designed to infer the quasi-periodic pattern of heartbeats, which eliminates temporal disturbances such as head movements. The two TokenLearners, S-TL and T-TL, are executed in a dual-path mode. This enables the model to reduce noise disturbances for final rPPG signal prediction. Extensive experiments on four physiological measurement benchmark datasets are conducted. The Dual-TL achieves state-of-the-art performances in both intra- and cross-dataset testings, demonstrating its immense potential as a basic backbone for rPPG measurement. The source code is available at \href{https://github.com/VUT-HFUT/Dual-TL}{https://github.com/VUT-HFUT/Dual-TL}

Abstract: Recently, Transformer-based architecture has been introduced into single image deraining task due to its advantage in modeling non-local information. However, existing approaches tend to integrate global features based on a dense self-attention strategy since it tend to uses all similarities of the tokens between the queries and keys. In fact, this strategy leads to ignoring the most relevant information and inducing blurry effect by the irrelevant representations during the feature aggregation. To this end, this paper proposes an effective image deraining Transformer with dynamic dual self-attention (DDSA), which combines both dense and sparse attention strategies to better facilitate clear image reconstruction. Specifically, we only select the most useful similarity values based on top-k approximate calculation to achieve sparse attention. In addition, we also develop a novel spatial-enhanced feed-forward network (SEFN) to further obtain a more accurate representation for achieving high-quality derained results. Extensive experiments on benchmark datasets demonstrate the effectiveness of our proposed method.

Abstract: Video anomaly detection (VAD) is an important but challenging task in computer vision. The main challenge rises due to the rarity of training samples to model all anomaly cases. Hence, semi-supervised anomaly detection methods have gotten more attention, since they focus on modeling normals and they detect anomalies by measuring the deviations from normal patterns. Despite impressive advances of these methods in modeling normal motion and appearance, long-term motion modeling has not been effectively explored so far. Inspired by the abilities of the future frame prediction proxy-task, we introduce the task of future video prediction from a single frame, as a novel proxy-task for video anomaly detection. This proxy-task alleviates the challenges of previous methods in learning longer motion patterns. Moreover, we replace the initial and future raw frames with their corresponding semantic segmentation map, which not only makes the method aware of object class but also makes the prediction task less complex for the model. Extensive experiments on the benchmark datasets (ShanghaiTech, UCSD-Ped1, and UCSD-Ped2) show the effectiveness of the method and the superiority of its performance compared to SOTA prediction-based VAD methods.

Abstract: Recent work on deep reinforcement learning (DRL) has pointed out that algorithmic information about good policies can be extracted from offline data which lack explicit information about executed actions. For example, videos of humans or robots may convey a lot of implicit information about rewarding action sequences, but a DRL machine that wants to profit from watching such videos must first learn by itself to identify and recognize relevant states/actions/rewards. Without relying on ground-truth annotations, our new method called Deep State Identifier learns to predict returns from episodes encoded as videos. Then it uses a kind of mask-based sensitivity analysis to extract/identify important critical states. Extensive experiments showcase our method's potential for understanding and improving agent behavior. The source code and the generated datasets are available at https://github.com/AI-Initiative-KAUST/VideoRLCS.

Abstract: Purpose: In this paper, we establish a baseline for handwritten stenography recognition, using the novel LION dataset, and investigate the impact of including selected aspects of stenographic theory into the recognition process. We make the LION dataset publicly available with the aim of encouraging future research in handwritten stenography recognition. Methods: A state-of-the-art text recognition model is trained to establish a baseline. Stenographic domain knowledge is integrated by applying four different encoding methods that transform the target sequence into representations, which approximate selected aspects of the writing system. Results are further improved by integrating a pre-training scheme, based on synthetic data. Results: The baseline model achieves an average test character error rate (CER) of 29.81% and a word error rate (WER) of 55.14%. Test error rates are reduced significantly by combining stenography-specific target sequence encodings with pre-training and fine-tuning, yielding CERs in the range of 24.5% - 26% and WERs of 44.8% - 48.2%. Conclusion: The obtained results demonstrate the challenging nature of stenography recognition. Integrating stenography-specific knowledge, in conjunction with pre-training and fine-tuning on synthetic data, yields considerable improvements. Together with our precursor study on the subject, this is the first work to apply modern handwritten text recognition to stenography. The dataset and our code are publicly available via Zenodo.

Abstract: Neuromorphic vision sensors, or event cameras, differ from conventional cameras in that they do not capture images at a specified rate. Instead, they asynchronously log local brightness changes at each pixel. As a result, event cameras only record changes in a given scene, and do so with very high temporal resolution, high dynamic range, and low power requirements. Recent research has demonstrated how these characteristics make event cameras extremely practical sensors in driver monitoring systems (DMS), enabling the tracking of high-speed eye motion and blinks. This research provides a proof of concept to expand event-based DMS techniques to include seatbelt state detection. Using an event simulator, a dataset of 108,691 synthetic neuromorphic frames of car occupants was generated from a near-infrared (NIR) dataset, and split into training, validation, and test sets for a seatbelt state detection algorithm based on a recurrent convolutional neural network (CNN). In addition, a smaller set of real event data was collected and reserved for testing. In a binary classification task, the fastened/unfastened frames were identified with an F1 score of 0.989 and 0.944 on the simulated and real test sets respectively. When the problem extended to also classify the action of fastening/unfastening the seatbelt, respective F1 scores of 0.964 and 0.846 were achieved.

Abstract: Class imbalance is a common challenge in real-world recognition tasks, where the majority of classes have few samples, also known as tail classes. We address this challenge with the perspective of generalization and empirically find that the promising Sharpness-Aware Minimization (SAM) fails to address generalization issues under the class-imbalanced setting. Through investigating this specific type of task, we identify that its generalization bottleneck primarily lies in the severe overfitting for tail classes with limited training data. To overcome this bottleneck, we leverage class priors to restrict the generalization scope of the class-agnostic SAM and propose a class-aware smoothness optimization algorithm named Imbalanced-SAM (ImbSAM). With the guidance of class priors, our ImbSAM specifically improves generalization targeting tail classes. We also verify the efficacy of ImbSAM on two prototypical applications of class-imbalanced recognition: long-tailed classification and semi-supervised anomaly detection, where our ImbSAM demonstrates remarkable performance improvements for tail classes and anomaly. Our code implementation is available at https://github.com/cool-xuan/Imbalanced_SAM.

Abstract: We investigate the impact of pre-defined keypoints for pose estimation, and found that accuracy and efficiency can be improved by training a graph network to select a set of disperse keypoints with similarly distributed votes. These votes, learned by a regression network to accumulate evidence for the keypoint locations, can be regressed more accurately compared to previous heuristic keypoint algorithms. The proposed KeyGNet, supervised by a combined loss measuring both Wassserstein distance and dispersion, learns the color and geometry features of the target objects to estimate optimal keypoint locations. Experiments demonstrate the keypoints selected by KeyGNet improved the accuracy for all evaluation metrics of all seven datasets tested, for three keypoint voting methods. The challenging Occlusion LINEMOD dataset notably improved ADD(S) by +16.4% on PVN3D, and all core BOP datasets showed an AR improvement for all objects, of between +1% and +21.5%. There was also a notable increase in performance when transitioning from single object to multiple object training using KeyGNet keypoints, essentially eliminating the SISO-MIMO gap for Occlusion LINEMOD.

Abstract: In this paper, we present a novel shape reconstruction method leveraging diffusion model to generate 3D sparse point cloud for the object captured in a single RGB image. Recent methods typically leverage global embedding or local projection-based features as the condition to guide the diffusion model. However, such strategies fail to consistently align the denoised point cloud with the given image, leading to unstable conditioning and inferior performance. In this paper, we present CCD-3DR, which exploits a novel centered diffusion probabilistic model for consistent local feature conditioning. We constrain the noise and sampled point cloud from the diffusion model into a subspace where the point cloud center remains unchanged during the forward diffusion process and reverse process. The stable point cloud center further serves as an anchor to align each point with its corresponding local projection-based features. Extensive experiments on synthetic benchmark ShapeNet-R2N2 demonstrate that CCD-3DR outperforms all competitors by a large margin, with over 40% improvement. We also provide results on real-world dataset Pix3D to thoroughly demonstrate the potential of CCD-3DR in real-world applications. Codes will be released soon

Abstract: Content and style (C-S) disentanglement is a fundamental problem and critical challenge of style transfer. Existing approaches based on explicit definitions (e.g., Gram matrix) or implicit learning (e.g., GANs) are neither interpretable nor easy to control, resulting in entangled representations and less satisfying results. In this paper, we propose a new C-S disentangled framework for style transfer without using previous assumptions. The key insight is to explicitly extract the content information and implicitly learn the complementary style information, yielding interpretable and controllable C-S disentanglement and style transfer. A simple yet effective CLIP-based style disentanglement loss coordinated with a style reconstruction prior is introduced to disentangle C-S in the CLIP image space. By further leveraging the powerful style removal and generative ability of diffusion models, our framework achieves superior results than state of the art and flexible C-S disentanglement and trade-off control. Our work provides new insights into the C-S disentanglement in style transfer and demonstrates the potential of diffusion models for learning well-disentangled C-S characteristics.

Abstract: In recent years, neural implicit surface reconstruction has emerged as a popular paradigm for multi-view 3D reconstruction. Unlike traditional multi-view stereo approaches, the neural implicit surface-based methods leverage neural networks to represent 3D scenes as signed distance functions (SDFs). However, they tend to disregard the reconstruction of individual objects within the scene, which limits their performance and practical applications. To address this issue, previous work ObjectSDF introduced a nice framework of object-composition neural implicit surfaces, which utilizes 2D instance masks to supervise individual object SDFs. In this paper, we propose a new framework called ObjectSDF++ to overcome the limitations of ObjectSDF. First, in contrast to ObjectSDF whose performance is primarily restricted by its converted semantic field, the core component of our model is an occlusion-aware object opacity rendering formulation that directly volume-renders object opacity to be supervised with instance masks. Second, we design a novel regularization term for object distinction, which can effectively mitigate the issue that ObjectSDF may result in unexpected reconstruction in invisible regions due to the lack of constraint to prevent collisions. Our extensive experiments demonstrate that our novel framework not only produces superior object reconstruction results but also significantly improves the quality of scene reconstruction. Code and more resources can be found in \url{https://qianyiwu.github.io/objectsdf++}

Abstract: Deep Learning methods have recently seen increased adoption in medical imaging applications. However, elevated vulnerabilities have been explored in recent Deep Learning solutions, which can hinder future adoption. Particularly, the vulnerability of Vision Transformer (ViT) to adversarial, privacy, and confidentiality attacks raise serious concerns about their reliability in medical settings. This work aims to enhance the robustness of self-ensembling ViTs for the tuberculosis chest x-ray classification task. We propose Self-Ensembling ViT with defensive Distillation and Adversarial training (SEDA). SEDA utilizes efficient CNN blocks to learn spatial features with various levels of abstraction from feature representations extracted from intermediate ViT blocks, that are largely unaffected by adversarial perturbations. Furthermore, SEDA leverages adversarial training in combination with defensive distillation for improved robustness against adversaries. Training using adversarial examples leads to better model generalizability and improves its ability to handle perturbations. Distillation using soft probabilities introduces uncertainty and variation into the output probabilities, making it more difficult for adversarial and privacy attacks. Extensive experiments performed with the proposed architecture and training paradigm on publicly available Tuberculosis x-ray dataset shows SOTA efficacy of SEDA compared to SEViT in terms of computational efficiency with 70x times lighter framework and enhanced robustness of +9%.

Abstract: The ability to learn from context with novel concepts, and deliver appropriate responses are essential in human conversations. Despite current Multimodal Large Language Models (MLLMs) and Large Language Models (LLMs) being trained on mega-scale datasets, recognizing unseen images or understanding novel concepts in a training-free manner remains a challenge. In-Context Learning (ICL) explores training-free few-shot learning, where models are encouraged to ``learn to learn" from limited tasks and generalize to unseen tasks. In this work, we propose link-context learning (LCL), which emphasizes "reasoning from cause and effect" to augment the learning capabilities of MLLMs. LCL goes beyond traditional ICL by explicitly strengthening the causal relationship between the support set and the query set. By providing demonstrations with causal links, LCL guides the model to discern not only the analogy but also the underlying causal associations between data points, which empowers MLLMs to recognize unseen images and understand novel concepts more effectively. To facilitate the evaluation of this novel approach, we introduce the ISEKAI dataset, comprising exclusively of unseen generated image-label pairs designed for link-context learning. Extensive experiments show that our LCL-MLLM exhibits strong link-context learning capabilities to novel concepts over vanilla MLLMs. Code and data will be released at https://github.com/isekai-portal/Link-Context-Learning.

Abstract: Most existing forecasting systems are memory-based methods, which attempt to mimic human forecasting ability by employing various memory mechanisms and have progressed in temporal modeling for memory dependency. Nevertheless, an obvious weakness of this paradigm is that it can only model limited historical dependence and can not transcend the past. In this paper, we rethink the temporal dependence of event evolution and propose a novel memory-anticipation-based paradigm to model an entire temporal structure, including the past, present, and future. Based on this idea, we present Memory-and-Anticipation Transformer (MAT), a memory-anticipation-based approach, to address the online action detection and anticipation tasks. In addition, owing to the inherent superiority of MAT, it can process online action detection and anticipation tasks in a unified manner. The proposed MAT model is tested on four challenging benchmarks TVSeries, THUMOS'14, HDD, and EPIC-Kitchens-100, for online action detection and anticipation tasks, and it significantly outperforms all existing methods. Code is available at https://github.com/Echo0125/Memory-and-Anticipation-Transformer.

Abstract: Foundation vision-language models are currently transforming computer vision, and are on the rise in medical imaging fueled by their very promising generalization capabilities. However, the initial attempts to transfer this new paradigm to medical imaging have shown less impressive performances than those observed in other domains, due to the significant domain shift and the complex, expert domain knowledge inherent to medical-imaging tasks. Motivated by the need for domain-expert foundation models, we present FLAIR, a pre-trained vision-language model for universal retinal fundus image understanding. To this end, we compiled 37 open-access, mostly categorical fundus imaging datasets from various sources, with up to 97 different target conditions and 284,660 images. We integrate the expert's domain knowledge in the form of descriptive textual prompts, during both pre-training and zero-shot inference, enhancing the less-informative categorical supervision of the data. Such a textual expert's knowledge, which we compiled from the relevant clinical literature and community standards, describes the fine-grained features of the pathologies as well as the hierarchies and dependencies between them. We report comprehensive evaluations, which illustrate the benefit of integrating expert knowledge and the strong generalization capabilities of FLAIR under difficult scenarios with domain shifts or unseen categories. When adapted with a lightweight linear probe, FLAIR outperforms fully-trained, dataset-focused models, more so in the few-shot regimes. Interestingly, FLAIR outperforms by a large margin more generalist, larger-scale image-language models, which emphasizes the potential of embedding experts' domain knowledge and the limitations of generalist models in medical imaging.

Abstract: This paper tackles the challenge of creating relightable and animatable neural avatars from sparse-view (or even monocular) videos of dynamic humans under unknown illumination. Compared to studio environments, this setting is more practical and accessible but poses an extremely challenging ill-posed problem. Previous neural human reconstruction methods are able to reconstruct animatable avatars from sparse views using deformed Signed Distance Fields (SDF) but cannot recover material parameters for relighting. While differentiable inverse rendering-based methods have succeeded in material recovery of static objects, it is not straightforward to extend them to dynamic humans as it is computationally intensive to compute pixel-surface intersection and light visibility on deformed SDFs for inverse rendering. To solve this challenge, we propose a Hierarchical Distance Query (HDQ) algorithm to approximate the world space distances under arbitrary human poses. Specifically, we estimate coarse distances based on a parametric human model and compute fine distances by exploiting the local deformation invariance of SDF. Based on the HDQ algorithm, we leverage sphere tracing to efficiently estimate the surface intersection and light visibility. This allows us to develop the first system to recover animatable and relightable neural avatars from sparse view (or monocular) inputs. Experiments demonstrate that our approach is able to produce superior results compared to state-of-the-art methods. Our code will be released for reproducibility.

Abstract: We introduce an object-aware decoder for improving the performance of spatio-temporal representations on ego-centric videos. The key idea is to enhance object-awareness during training by tasking the model to predict hand positions, object positions, and the semantic label of the objects using paired captions when available. At inference time the model only requires RGB frames as inputs, and is able to track and ground objects (although it has not been trained explicitly for this). We demonstrate the performance of the object-aware representations learnt by our model, by: (i) evaluating it for strong transfer, i.e. through zero-shot testing, on a number of downstream video-text retrieval and classification benchmarks; and (ii) by using the representations learned as input for long-term video understanding tasks (e.g. Episodic Memory in Ego4D). In all cases the performance improves over the state of the art -- even compared to networks trained with far larger batch sizes. We also show that by using noisy image-level detection as pseudo-labels in training, the model learns to provide better bounding boxes using video consistency, as well as grounding the words in the associated text descriptions. Overall, we show that the model can act as a drop-in replacement for an ego-centric video model to improve performance through visual-text grounding.

Abstract: We present the content deformation field CoDeF as a new type of video representation, which consists of a canonical content field aggregating the static contents in the entire video and a temporal deformation field recording the transformations from the canonical image (i.e., rendered from the canonical content field) to each individual frame along the time axis.Given a target video, these two fields are jointly optimized to reconstruct it through a carefully tailored rendering pipeline.We advisedly introduce some regularizations into the optimization process, urging the canonical content field to inherit semantics (e.g., the object shape) from the video.With such a design, CoDeF naturally supports lifting image algorithms for video processing, in the sense that one can apply an image algorithm to the canonical image and effortlessly propagate the outcomes to the entire video with the aid of the temporal deformation field.We experimentally show that CoDeF is able to lift image-to-image translation to video-to-video translation and lift keypoint detection to keypoint tracking without any training.More importantly, thanks to our lifting strategy that deploys the algorithms on only one image, we achieve superior cross-frame consistency in processed videos compared to existing video-to-video translation approaches, and even manage to track non-rigid objects like water and smog.Project page can be found at https://qiuyu96.github.io/CoDeF/.