Computer Vision and Pattern Recognition (cs.CV)

Abstract: One of the most challenging problems in audio-driven talking head generation is achieving high-fidelity detail while ensuring precise synchronization. Given only a single reference image, extracting meaningful identity attributes becomes even more challenging, often causing the network to mirror the facial and lip structures too closely. To address these issues, we introduce RADIO, a framework engineered to yield high-quality dubbed videos regardless of the pose or expression in reference images. The key is to modulate the decoder layers using latent space composed of audio and reference features. Additionally, we incorporate ViT blocks into the decoder to emphasize high-fidelity details, especially in the lip region. Our experimental results demonstrate that RADIO displays high synchronization without the loss of fidelity. Especially in harsh scenarios where the reference frame deviates significantly from the ground truth, our method outperforms state-of-the-art methods, highlighting its robustness. Pre-trained model and codes will be made public after the review.

Abstract: Deep neural networks have achieved remarkable progress in enhancing low-light images by improving their brightness and eliminating noise. However, most existing methods construct end-to-end mapping networks heuristically, neglecting the intrinsic prior of image enhancement task and lacking transparency and interpretability. Although some unfolding solutions have been proposed to relieve these issues, they rely on proximal operator networks that deliver ambiguous and implicit priors. In this work, we propose a paradigm for low-light image enhancement that explores the potential of customized learnable priors to improve the transparency of the deep unfolding paradigm. Motivated by the powerful feature representation capability of Masked Autoencoder (MAE), we customize MAE-based illumination and noise priors and redevelop them from two perspectives: 1) \textbf{structure flow}: we train the MAE from a normal-light image to its illumination properties and then embed it into the proximal operator design of the unfolding architecture; and m2) \textbf{optimization flow}: we train MAE from a normal-light image to its gradient representation and then employ it as a regularization term to constrain noise in the model output. These designs improve the interpretability and representation capability of the model.Extensive experiments on multiple low-light image enhancement datasets demonstrate the superiority of our proposed paradigm over state-of-the-art methods. Code is available at https://github.com/zheng980629/CUE.

Abstract: In this report, we introduce NICE project\footnote{\url{https://nice.lgresearch.ai/}} and share the results and outcomes of NICE challenge 2023. This project is designed to challenge the computer vision community to develop robust image captioning models that advance the state-of-the-art both in terms of accuracy and fairness. Through the challenge, the image captioning models were tested using a new evaluation dataset that includes a large variety of visual concepts from many domains. There was no specific training data provided for the challenge, and therefore the challenge entries were required to adapt to new types of image descriptions that had not been seen during training. This report includes information on the newly proposed NICE dataset, evaluation methods, challenge results, and technical details of top-ranking entries. We expect that the outcomes of the challenge will contribute to the improvement of AI models on various vision-language tasks.

Abstract: Using additional training data is known to improve the results, especially for medical image 3D segmentation where there is a lack of training material and the model needs to generalize well from few available data. However, the new data could have been acquired using other instruments and preprocessed such its distribution is significantly different from the original training data. Therefore, we study techniques which ameliorate domain shift during training so that the additional data becomes better usable for preprocessing and training together with the original data. Our results show that transforming the additional data using histogram matching has better results than using simple normalization.

Abstract: Vision Transformer (ViT) architectures are becoming increasingly popular and widely employed to tackle computer vision applications. Their main feature is the capacity to extract global information through the self-attention mechanism, outperforming earlier convolutional neural networks. However, ViT deployment and performance have grown steadily with their size, number of trainable parameters, and operations. Furthermore, self-attention's computational and memory cost quadratically increases with the image resolution. Generally speaking, it is challenging to employ these architectures in real-world applications due to many hardware and environmental restrictions, such as processing and computational capabilities. Therefore, this survey investigates the most efficient methodologies to ensure sub-optimal estimation performances. More in detail, four efficient categories will be analyzed: compact architecture, pruning, knowledge distillation, and quantization strategies. Moreover, a new metric called Efficient Error Rate has been introduced in order to normalize and compare models' features that affect hardware devices at inference time, such as the number of parameters, bits, FLOPs, and model size. Summarizing, this paper firstly mathematically defines the strategies used to make Vision Transformer efficient, describes and discusses state-of-the-art methodologies, and analyzes their performances over different application scenarios. Toward the end of this paper, we also discuss open challenges and promising research directions.

Abstract: Referring video object segmentation (RVOS), as a supervised learning task, relies on sufficient annotated data for a given scene. However, in more realistic scenarios, only minimal annotations are available for a new scene, which poses significant challenges to existing RVOS methods. With this in mind, we propose a simple yet effective model with a newly designed cross-modal affinity (CMA) module based on a Transformer architecture. The CMA module builds multimodal affinity with a few samples, thus quickly learning new semantic information, and enabling the model to adapt to different scenarios. Since the proposed method targets limited samples for new scenes, we generalize the problem as - few-shot referring video object segmentation (FS-RVOS). To foster research in this direction, we build up a new FS-RVOS benchmark based on currently available datasets. The benchmark covers a wide range and includes multiple situations, which can maximally simulate real-world scenarios. Extensive experiments show that our model adapts well to different scenarios with only a few samples, reaching state-of-the-art performance on the benchmark. On Mini-Ref-YouTube-VOS, our model achieves an average performance of 53.1 J and 54.8 F, which are 10% better than the baselines. Furthermore, we show impressive results of 77.7 J and 74.8 F on Mini-Ref-SAIL-VOS, which are significantly better than the baselines. Code is publicly available at https://github.com/hengliusky/Few_shot_RVOS.

Abstract: RGB-guided depth completion aims at predicting dense depth maps from sparse depth measurements and corresponding RGB images, where how to effectively and efficiently exploit the multi-modal information is a key issue. Guided dynamic filters, which generate spatially-variant depth-wise separable convolutional filters from RGB features to guide depth features, have been proven to be effective in this task. However, the dynamically generated filters require massive model parameters, computational costs and memory footprints when the number of feature channels is large. In this paper, we propose to decompose the guided dynamic filters into a spatially-shared component multiplied by content-adaptive adaptors at each spatial location. Based on the proposed idea, we introduce two decomposition schemes A and B, which decompose the filters by splitting the filter structure and using spatial-wise attention, respectively. The decomposed filters not only maintain the favorable properties of guided dynamic filters as being content-dependent and spatially-variant, but also reduce model parameters and hardware costs, as the learned adaptors are decoupled with the number of feature channels. Extensive experimental results demonstrate that the methods using our schemes outperform state-of-the-art methods on the KITTI dataset, and rank 1st and 2nd on the KITTI benchmark at the time of submission. Meanwhile, they also achieve comparable performance on the NYUv2 dataset. In addition, our proposed methods are general and could be employed as plug-and-play feature fusion blocks in other multi-modal fusion tasks such as RGB-D salient object detection.

Abstract: 3D object detection is an essential task for achieving autonomous driving. Existing anchor-based detection methods rely on empirical heuristics setting of anchors, which makes the algorithms lack elegance. In recent years, we have witnessed the rise of several generative models, among which diffusion models show great potential for learning the transformation of two distributions. Our proposed Diff3Det migrates the diffusion model to proposal generation for 3D object detection by considering the detection boxes as generative targets. During training, the object boxes diffuse from the ground truth boxes to the Gaussian distribution, and the decoder learns to reverse this noise process. In the inference stage, the model progressively refines a set of random boxes to the prediction results. We provide detailed experiments on the KITTI benchmark and achieve promising performance compared to classical anchor-based 3D detection methods.

Abstract: Detecting small moving targets accurately in infrared (IR) image sequences is a significant challenge. To address this problem, we propose a novel method called spatial-temporal local feature difference (STLFD) with adaptive background suppression (ABS). Our approach utilizes filters in the spatial and temporal domains and performs pixel-level ABS on the output to enhance the contrast between the target and the background. The proposed method comprises three steps. First, we obtain three temporal frame images based on the current frame image and extract two feature maps using the designed spatial domain and temporal domain filters. Next, we fuse the information of the spatial domain and temporal domain to produce the spatial-temporal feature maps and suppress noise using our pixel-level ABS module. Finally, we obtain the segmented binary map by applying a threshold. Our experimental results demonstrate that the proposed method outperforms existing state-of-the-art methods for infrared small-moving target detection.

Abstract: A set of features independent of character stroke direction and order variations is proposed for online handwritten character recognition. A method is developed that maps features like co-ordinates of points, orientations of strokes at points, and dynamics of orientations of strokes at points spatially as a function of co-ordinate values of the points and computes histograms of these features from different regions in the spatial map. Different features like spatio-temporal, discrete Fourier transform, discrete cosine transform, discrete wavelet transform, spatial, and histograms of oriented gradients used in other studies for training classifiers for character recognition are considered. The classifier chosen for classification performance comparison, when trained with different features, is support vector machines (SVM). The character datasets used for training and testing the classifiers consist of online handwritten samples of 96 different Hindi characters. There are 12832 and 2821 samples in training and testing datasets, respectively. SVM classifiers trained with the proposed features has the highest classification accuracy of 92.9\% when compared to the performances of SVM classifiers trained with the other features and tested on the same testing dataset. Therefore, the proposed features have better character discriminative capability than the other features considered for comparison.

Abstract: Support-query shift few-shot learning aims to classify unseen examples (query set) to labeled data (support set) based on the learned embedding in a low-dimensional space under a distribution shift between the support set and the query set. However, in real-world scenarios the shifts are usually unknown and varied, making it difficult to estimate in advance. Therefore, in this paper, we propose a novel but more difficult challenge, RSQS, focusing on Realistic Support-Query Shift few-shot learning. The key feature of RSQS is that the individual samples in a meta-task are subjected to multiple distribution shifts in each meta-task. In addition, we propose a unified adversarial feature alignment method called DUal adversarial ALignment framework (DuaL) to relieve RSQS from two aspects, i.e., inter-domain bias and intra-domain variance. On the one hand, for the inter-domain bias, we corrupt the original data in advance and use the synthesized perturbed inputs to train the repairer network by minimizing distance in the feature level. On the other hand, for intra-domain variance, we proposed a generator network to synthesize hard, i.e., less similar, examples from the support set in a self-supervised manner and introduce regularized optimal transportation to derive a smooth optimal transportation plan. Lastly, a benchmark of RSQS is built with several state-of-the-art baselines among three datasets (CIFAR100, mini-ImageNet, and Tiered-Imagenet). Experiment results show that DuaL significantly outperforms the state-of-the-art methods in our benchmark.

Abstract: This paper presents Deep Networks for Improved Segmentation Edges (DeNISE), a novel data enhancement technique using edge detection and segmentation models to improve the boundary quality of segmentation masks. DeNISE utilizes the inherent differences in two sequential deep neural architectures to improve the accuracy of the predicted segmentation edge. DeNISE applies to all types of neural networks and is not trained end-to-end, allowing rapid experiments to discover which models complement each other. We test and apply DeNISE for building segmentation in aerial images. Aerial images are known for difficult conditions as they have a low resolution with optical noise, such as reflections, shadows, and visual obstructions. Overall the paper demonstrates the potential for DeNISE. Using the technique, we improve the baseline results with a building IoU of 78.9%.

Abstract: In this work, we consider the typography generation task that aims at producing diverse typographic styling for the given graphic document. We formulate typography generation as a fine-grained attribute generation for multiple text elements and build an autoregressive model to generate diverse typography that matches the input design context. We further propose a simple yet effective sampling approach that respects the consistency and distinction principle of typography so that generated examples share consistent typographic styling across text elements. Our empirical study shows that our model successfully generates diverse typographic designs while preserving a consistent typographic structure.

Abstract: Humans are good at recomposing novel objects, i.e. they can identify commonalities between unknown objects from general structure to finer detail, an ability difficult to replicate by machines. We propose a framework, ISCO, to recompose an object using 3D superquadrics as semantic parts directly from 2D views without training a model that uses 3D supervision. To achieve this, we optimize the superquadric parameters that compose a specific instance of the object, comparing its rendered 3D view and 2D image silhouette. Our ISCO framework iteratively adds new superquadrics wherever the reconstruction error is high, abstracting first coarse regions and then finer details of the target object. With this simple coarse-to-fine inductive bias, ISCO provides consistent superquadrics for related object parts, despite not having any semantic supervision. Since ISCO does not train any neural network, it is also inherently robust to out-of-distribution objects. Experiments show that, compared to recent single instance superquadrics reconstruction approaches, ISCO provides consistently more accurate 3D reconstructions, even from images in the wild. Code available at https://github.com/ExplainableML/ISCO .

Abstract: Video outpainting aims to adequately complete missing areas at the edges of video frames. Compared to image outpainting, it presents an additional challenge as the model should maintain the temporal consistency of the filled area. In this paper, we introduce a masked 3D diffusion model for video outpainting. We use the technique of mask modeling to train the 3D diffusion model. This allows us to use multiple guide frames to connect the results of multiple video clip inferences, thus ensuring temporal consistency and reducing jitter between adjacent frames. Meanwhile, we extract the global frames of the video as prompts and guide the model to obtain information other than the current video clip using cross-attention. We also introduce a hybrid coarse-to-fine inference pipeline to alleviate the artifact accumulation problem. The existing coarse-to-fine pipeline only uses the infilling strategy, which brings degradation because the time interval of the sparse frames is too large. Our pipeline benefits from bidirectional learning of the mask modeling and thus can employ a hybrid strategy of infilling and interpolation when generating sparse frames. Experiments show that our method achieves state-of-the-art results in video outpainting tasks. More results are provided at our https://fanfanda.github.io/M3DDM/.

Abstract: Accurate affordance detection and segmentation with pixel precision is an important piece in many complex systems based on interactions, such as robots and assitive devices. We present a new approach to affordance perception which enables accurate multi-label segmentation. Our approach can be used to automatically extract grounded affordances from first person videos of interactions using a 3D map of the environment providing pixel level precision for the affordance location. We use this method to build the largest and most complete dataset on affordances based on the EPIC-Kitchen dataset, EPIC-Aff, which provides interaction-grounded, multi-label, metric and spatial affordance annotations. Then, we propose a new approach to affordance segmentation based on multi-label detection which enables multiple affordances to co-exists in the same space, for example if they are associated with the same object. We present several strategies of multi-label detection using several segmentation architectures. The experimental results highlight the importance of the multi-label detection. Finally, we show how our metric representation can be exploited for build a map of interaction hotspots in spatial action-centric zones and use that representation to perform a task-oriented navigation.

Abstract: Airborne LiDAR systems have the capability to capture the Earth's surface by generating extensive point cloud data comprised of points mainly defined by 3D coordinates. However, labeling such points for supervised learning tasks is time-consuming. As a result, there is a need to investigate techniques that can learn from unlabeled data to significantly reduce the number of annotated samples. In this work, we propose to train a self-supervised encoder with Barlow Twins and use it as a pre-trained network in the task of semantic scene segmentation. The experimental results demonstrate that our unsupervised pre-training boosts performance once fine-tuned on the supervised task, especially for under-represented categories.

Abstract: The advent of satellite-borne machine learning hardware accelerators has enabled the on-board processing of payload data using machine learning techniques such as convolutional neural networks (CNN). A notable example is using a CNN to detect the presence of clouds in hyperspectral data captured on Earth observation (EO) missions, whereby only clear sky data is downlinked to conserve bandwidth. However, prior to deployment, new missions that employ new sensors will not have enough representative datasets to train a CNN model, while a model trained solely on data from previous missions will underperform when deployed to process the data on the new missions. This underperformance stems from the domain gap, i.e., differences in the underlying distributions of the data generated by the different sensors in previous and future missions. In this paper, we address the domain gap problem in the context of on-board hyperspectral cloud detection. Our main contributions lie in formulating new domain adaptation tasks that are motivated by a concrete EO mission, developing a novel algorithm for bandwidth-efficient supervised domain adaptation, and demonstrating test-time adaptation algorithms on space deployable neural network accelerators. Our contributions enable minimal data transmission to be invoked (e.g., only 1% of the weights in ResNet50) to achieve domain adaptation, thereby allowing more sophisticated CNN models to be deployed and updated on satellites without being hampered by domain gap and bandwidth limitations.

Abstract: VQA Natural Language Explanation (VQA-NLE) task aims to explain the decision-making process of VQA models in natural language. Unlike traditional attention or gradient analysis, free-text rationales can be easier to understand and gain users' trust. Existing methods mostly use post-hoc or self-rationalization models to obtain a plausible explanation. However, these frameworks are bottlenecked by the following challenges: 1) the reasoning process cannot be faithfully responded to and suffer from the problem of logical inconsistency. 2) Human-annotated explanations are expensive and time-consuming to collect. In this paper, we propose a new Semi-Supervised VQA-NLE via Self-Critical Learning (S3C), which evaluates the candidate explanations by answering rewards to improve the logical consistency between answers and rationales. With a semi-supervised learning framework, the S3C can benefit from a tremendous amount of samples without human-annotated explanations. A large number of automatic measures and human evaluations all show the effectiveness of our method. Meanwhile, the framework achieves a new state-of-the-art performance on the two VQA-NLE datasets.

Abstract: Real-time traffic light recognition is essential for autonomous driving. Yet, a cohesive overview of the underlying model architectures for this task is currently missing. In this work, we conduct a comprehensive survey and analysis of traffic light recognition methods that use convolutional neural networks (CNNs). We focus on two essential aspects: datasets and CNN architectures. Based on an underlying architecture, we cluster methods into three major groups: (1) modifications of generic object detectors which compensate for specific task characteristics, (2) multi-stage approaches involving both rule-based and CNN components, and (3) task-specific single-stage methods. We describe the most important works in each cluster, discuss the usage of the datasets, and identify research gaps.

Abstract: Although recent deep learning based gaze estimation approaches have achieved much improvement, we still know little about how gaze features are connected to the physics of gaze. In this paper, we try to answer this question by analyzing the gaze feature manifold. Our analysis revealed the insight that the geodesic distance between gaze features is consistent with the gaze differences between samples. According to this finding, we construct the Physics- Consistent Feature (PCF) in an analytical way, which connects gaze feature to the physical definition of gaze. We further propose the PCFGaze framework that directly optimizes gaze feature space by the guidance of PCF. Experimental results demonstrate that the proposed framework alleviates the overfitting problem and significantly improves cross-domain gaze estimation accuracy without extra training data. The insight of gaze feature has the potential to benefit other regression tasks with physical meanings.

Abstract: Composed image retrieval, a task involving the search for a target image using a reference image and a complementary text as the query, has witnessed significant advancements owing to the progress made in cross-modal modeling. Unlike the general image-text retrieval problem with only one alignment relation, i.e., image-text, we argue for the existence of two types of relations in composed image retrieval. The explicit relation pertains to the reference image & complementary text-target image, which is commonly exploited by existing methods. Besides this intuitive relation, the observations during our practice have uncovered another implicit yet crucial relation, i.e., reference image & target image-complementary text, since we found that the complementary text can be inferred by studying the relation between the target image and the reference image. Regrettably, existing methods largely focus on leveraging the explicit relation to learn their networks, while overlooking the implicit relation. In response to this weakness, We propose a new framework for composed image retrieval, termed dual relation alignment, which integrates both explicit and implicit relations to fully exploit the correlations among the triplets. Specifically, we design a vision compositor to fuse reference image and target image at first, then the resulted representation will serve two roles: (1) counterpart for semantic alignment with the complementary text and (2) compensation for the complementary text to boost the explicit relation modeling, thereby implant the implicit relation into the alignment learning. Our method is evaluated on two popular datasets, CIRR and FashionIQ, through extensive experiments. The results confirm the effectiveness of our dual-relation learning in substantially enhancing composed image retrieval performance.

Abstract: 3D single object tracking (SOT) in point clouds is still a challenging problem due to appearance variation, distractors, and high sparsity of point clouds. Notably, in autonomous driving scenarios, the target object typically maintains spatial adjacency across consecutive frames, predominantly moving horizontally. This spatial continuity offers valuable prior knowledge for target localization. However, existing trackers, which often employ point-wise representations, struggle to efficiently utilize this knowledge owing to the irregular format of such representations. Consequently, they require elaborate designs and solving multiple subtasks to establish spatial correspondence. In this paper, we introduce BEVTrack, a simple yet strong baseline framework for 3D SOT. After converting consecutive point clouds into the common Bird's-Eye-View representation, BEVTrack inherently encodes spatial proximity and adeptly captures motion cues for tracking via a simple element-wise operation and convolutional layers. Additionally, to better deal with objects having diverse sizes and moving patterns, BEVTrack directly learns the underlying motion distribution rather than making a fixed Laplacian or Gaussian assumption as in previous works. Without bells and whistles, BEVTrack achieves state-of-the-art performance on KITTI and NuScenes datasets while maintaining a high inference speed of 122 FPS. The code will be released at https://github.com/xmm-prio/BEVTrack.

Abstract: Previous animatable 3D-aware GANs for human generation have primarily focused on either the human head or full body. However, head-only videos are relatively uncommon in real life, and full body generation typically does not deal with facial expression control and still has challenges in generating high-quality results. Towards applicable video avatars, we present an animatable 3D-aware GAN that generates portrait images with controllable facial expression, head pose, and shoulder movements. It is a generative model trained on unstructured 2D image collections without using 3D or video data. For the new task, we base our method on the generative radiance manifold representation and equip it with learnable facial and head-shoulder deformations. A dual-camera rendering and adversarial learning scheme is proposed to improve the quality of the generated faces, which is critical for portrait images. A pose deformation processing network is developed to generate plausible deformations for challenging regions such as long hair. Experiments show that our method, trained on unstructured 2D images, can generate diverse and high-quality 3D portraits with desired control over different properties.

Abstract: We study the problem of multimodal fusion in this paper. Recent exchanging-based methods have been proposed for vision-vision fusion, which aim to exchange embeddings learned from one modality to the other. However, most of them project inputs of multimodalities into different low-dimensional spaces and cannot be applied to the sequential input data. To solve these issues, in this paper, we propose a novel exchanging-based multimodal fusion model MuSE for text-vision fusion based on Transformer. We first use two encoders to separately map multimodal inputs into different low-dimensional spaces. Then we employ two decoders to regularize the embeddings and pull them into the same space. The two decoders capture the correlations between texts and images with the image captioning task and the text-to-image generation task, respectively. Further, based on the regularized embeddings, we present CrossTransformer, which uses two Transformer encoders with shared parameters as the backbone model to exchange knowledge between multimodalities. Specifically, CrossTransformer first learns the global contextual information of the inputs in the shallow layers. After that, it performs inter-modal exchange by selecting a proportion of tokens in one modality and replacing their embeddings with the average of embeddings in the other modality. We conduct extensive experiments to evaluate the performance of MuSE on the Multimodal Named Entity Recognition task and the Multimodal Sentiment Analysis task. Our results show the superiority of MuSE against other competitors. Our code and data are provided at https://github.com/RecklessRonan/MuSE.

Abstract: In many recent years, multi-view mammogram analysis has been focused widely on AI-based cancer assessment. In this work, we aim to explore diverse fusion strategies (average and concatenate) and examine the model's learning behavior with varying individuals and fusion pathways, involving Coarse Layer and Fine Layer. The Ipsilateral Multi-View Network, comprising five fusion types (Pre, Early, Middle, Last, and Post Fusion) in ResNet-18, is employed. Notably, the Middle Fusion emerges as the most balanced and effective approach, enhancing deep-learning models' generalization performance by +5.29\% (concatenate) and +5.9\% (average) in VinDr-Mammo dataset and +2.03\% (concatenate) and +3\% (average) in CMMD dataset on macro F1-Score. The paper emphasizes the crucial role of layer assignment in multi-view network extraction with various strategies.

Abstract: Accurate road surface classification is crucial for autonomous vehicles (AVs) to optimize driving conditions, enhance safety, and enable advanced road mapping. However, deep learning models for road surface classification suffer from poor generalization when tested on unseen datasets. To update these models with new information, also the original training dataset must be taken into account, in order to avoid catastrophic forgetting. This is, however, inefficient if not impossible, e.g., when the data is collected in streams or large amounts. To overcome this limitation and enable fast and efficient cross-dataset adaptation, we propose to employ continual learning finetuning methods designed to retain past knowledge while adapting to new data, thus effectively avoiding forgetting. Experimental results demonstrate the superiority of this approach over naive finetuning, achieving performance close to fresh retraining. While solving this known problem, we also provide a general description of how the same technique can be adopted in other AV scenarios. We highlight the potential computational and economic benefits that a continual-based adaptation can bring to the AV industry, while also reducing greenhouse emissions due to unnecessary joint retraining.

Abstract: Underwater image restoration has been a challenging problem for decades since the advent of underwater photography. Most solutions focus on shallow water scenarios, where the scene is uniformly illuminated by the sunlight. However, the vast majority of uncharted underwater terrain is located beyond 200 meters depth where natural light is scarce and artificial illumination is needed. In such cases, light sources co-moving with the camera, dynamically change the scene appearance, which make shallow water restoration methods inadequate. In particular for multi-light source systems (composed of dozens of LEDs nowadays), calibrating each light is time-consuming, error-prone and tedious, and we observe that only the integrated illumination within the viewing volume of the camera is critical, rather than the individual light sources. The key idea of this paper is therefore to exploit the appearance changes of objects or the seafloor, when traversing the viewing frustum of the camera. Through new constraints assuming Lambertian surfaces, corresponding image pixels constrain the light field in front of the camera, and for each voxel a signal factor and a backscatter value are stored in a volumetric grid that can be used for very efficient image restoration of camera-light platforms, which facilitates consistently texturing large 3D models and maps that would otherwise be dominated by lighting and medium artifacts. To validate the effectiveness of our approach, we conducted extensive experiments on simulated and real-world datasets. The results of these experiments demonstrate the robustness of our approach in restoring the true albedo of objects, while mitigating the influence of lighting and medium effects. Furthermore, we demonstrate our approach can be readily extended to other scenarios, including in-air imaging with artificial illumination or other similar cases.

Abstract: The rise of deepfake images, especially of well-known personalities, poses a serious threat to the dissemination of authentic information. To tackle this, we present a thorough investigation into how deepfakes are produced and how they can be identified. The cornerstone of our research is a rich collection of artificial celebrity faces, titled DeepFakeFace (DFF). We crafted the DFF dataset using advanced diffusion models and have shared it with the community through online platforms. This data serves as a robust foundation to train and test algorithms designed to spot deepfakes. We carried out a thorough review of the DFF dataset and suggest two evaluation methods to gauge the strength and adaptability of deepfake recognition tools. The first method tests whether an algorithm trained on one type of fake images can recognize those produced by other methods. The second evaluates the algorithm's performance with imperfect images, like those that are blurry, of low quality, or compressed. Given varied results across deepfake methods and image changes, our findings stress the need for better deepfake detectors. Our DFF dataset and tests aim to boost the development of more effective tools against deepfakes.

Abstract: Localizing objects in 3D scenes according to the semantics of a given natural language is a fundamental yet important task in the field of multimedia understanding, which benefits various real-world applications such as robotics and autonomous driving. However, the majority of existing 3D object grounding methods are restricted to a single-sentence input describing an individual object, which cannot comprehend and reason more contextualized descriptions of multiple objects in more practical 3D cases. To this end, we introduce a new challenging task, called 3D Dense Object Grounding (3D DOG), to jointly localize multiple objects described in a more complicated paragraph rather than a single sentence. Instead of naively localizing each sentence-guided object independently, we found that dense objects described in the same paragraph are often semantically related and spatially located in a focused region of the 3D scene. To explore such semantic and spatial relationships of densely referred objects for more accurate localization, we propose a novel Stacked Transformer based framework for 3D DOG, named 3DOGSFormer. Specifically, we first devise a contextual query-driven local transformer decoder to generate initial grounding proposals for each target object. Then, we employ a proposal-guided global transformer decoder that exploits the local object features to learn their correlation for further refining initial grounding proposals. Extensive experiments on three challenging benchmarks (Nr3D, Sr3D, and ScanRefer) show that our proposed 3DOGSFormer outperforms state-of-the-art 3D single-object grounding methods and their dense-object variants by significant margins.

Abstract: Onboard intelligent processing is widely applied in emergency tasks in the field of remote sensing. However, it is predominantly confined to an individual platform with a limited observation range as well as susceptibility to interference, resulting in limited accuracy. Considering the current state of multi-platform collaborative observation, this article innovatively presents a distributed collaborative perception network called DCP-Net. Firstly, the proposed DCP-Net helps members to enhance perception performance by integrating features from other platforms. Secondly, a self-mutual information match module is proposed to identify collaboration opportunities and select suitable partners, prioritizing critical collaborative features and reducing redundant transmission cost. Thirdly, a related feature fusion module is designed to address the misalignment between local and collaborative features, improving the quality of fused features for the downstream task. We conduct extensive experiments and visualization analyses using three semantic segmentation datasets, including Potsdam, iSAID and DFC23. The results demonstrate that DCP-Net outperforms the existing methods comprehensively, improving mIoU by 2.61%~16.89% at the highest collaboration efficiency, which promotes the performance to a state-of-the-art level.

Abstract: The advancement of machine learning algorithms in medical image analysis requires the expansion of training datasets. A popular and cost-effective approach is automated annotation extraction from free-text medical reports, primarily due to the high costs associated with expert clinicians annotating chest X-ray images. However, it has been shown that the resulting datasets are susceptible to biases and shortcuts. Another strategy to increase the size of a dataset is crowdsourcing, a widely adopted practice in general computer vision with some success in medical image analysis. In a similar vein to crowdsourcing, we enhance two publicly available chest X-ray datasets by incorporating non-expert annotations. However, instead of using diagnostic labels, we annotate shortcuts in the form of tubes. We collect 3.5k chest drain annotations for CXR14, and 1k annotations for 4 different tube types in PadChest. We train a chest drain detector with the non-expert annotations that generalizes well to expert labels. Moreover, we compare our annotations to those provided by experts and show "moderate" to "almost perfect" agreement. Finally, we present a pathology agreement study to raise awareness about ground truth annotations. We make our annotations and code available.

Abstract: Image deblurring is a critical task in the field of image restoration, aiming to eliminate blurring artifacts. However, the challenge of addressing non-uniform blurring leads to an ill-posed problem, which limits the generalization performance of existing deblurring models. To solve the problem, we propose a framework SAM-Deblur, integrating prior knowledge from the Segment Anything Model (SAM) into the deblurring task for the first time. In particular, SAM-Deblur is divided into three stages. First, We preprocess the blurred images, obtain image masks via SAM, and propose a mask dropout method for training to enhance model robustness. Then, to fully leverage the structural priors generated by SAM, we propose a Mask Average Pooling (MAP) unit specifically designed to average SAM-generated segmented areas, serving as a plug-and-play component which can be seamlessly integrated into existing deblurring networks. Finally, we feed the fused features generated by the MAP Unit into the deblurring model to obtain a sharp image. Experimental results on the RealBlurJ, ReloBlur, and REDS datasets reveal that incorporating our methods improves NAFNet's PSNR by 0.05, 0.96, and 7.03, respectively. Code will be available at \href{https://github.com/HPLQAQ/SAM-Deblur}{SAM-Deblur}.

Abstract: Current scene graph datasets suffer from strong long-tail distributions of their predicate classes. Due to a very low number of some predicate classes in the test sets, no reliable metrics can be retrieved for the rarest classes. We construct a new panoptic scene graph dataset and a set of metrics that are designed as a benchmark for the predictive performance especially on rare predicate classes. To construct the new dataset, we propose a model-assisted annotation pipeline that efficiently finds rare predicate classes that are hidden in a large set of images like needles in a haystack. Contrary to prior scene graph datasets, Haystack contains explicit negative annotations, i.e. annotations that a given relation does not have a certain predicate class. Negative annotations are helpful especially in the field of scene graph generation and open up a whole new set of possibilities to improve current scene graph generation models. Haystack is 100% compatible with existing panoptic scene graph datasets and can easily be integrated with existing evaluation pipelines. Our dataset and code can be found here: https://lorjul.github.io/haystack/. It includes annotation files and simple to use scripts and utilities, to help with integrating our dataset in existing work.

Abstract: Despite significant progress in video question answering (VideoQA), existing methods fall short of questions that require causal/temporal reasoning across frames. This can be attributed to imprecise motion representations. We introduce Action Temporality Modeling (ATM) for temporality reasoning via three-fold uniqueness: (1) rethinking the optical flow and realizing that optical flow is effective in capturing the long horizon temporality reasoning; (2) training the visual-text embedding by contrastive learning in an action-centric manner, leading to better action representations in both vision and text modalities; and (3) preventing the model from answering the question given the shuffled video in the fine-tuning stage, to avoid spurious correlation between appearance and motion and hence ensure faithful temporality reasoning. In the experiments, we show that ATM outperforms previous approaches in terms of the accuracy on multiple VideoQAs and exhibits better true temporality reasoning ability.

Abstract: Nowadays, the research on Large Vision-Language Models (LVLMs) has been significantly promoted thanks to the success of Large Language Models (LLM). Nevertheless, these Vision-Language Models (VLMs) are suffering from the drawback of hallucination -- due to insufficient understanding of vision and language modalities, VLMs may generate incorrect perception information when doing downstream applications, for example, captioning a non-existent entity. To address the hallucination phenomenon, on the one hand, we introduce a Contrastive Instruction Evaluation Method (CIEM), which is an automatic pipeline that leverages an annotated image-text dataset coupled with an LLM to generate factual/contrastive question-answer pairs for the evaluation of the hallucination of VLMs. On the other hand, based on CIEM, we further propose a new instruction tuning method called CIT (the abbreviation of Contrastive Instruction Tuning) to alleviate the hallucination of VLMs by automatically producing high-quality factual/contrastive question-answer pairs and corresponding justifications for model tuning. Through extensive experiments on CIEM and CIT, we pinpoint the hallucination issues commonly present in existing VLMs, the disability of the current instruction-tuning dataset to handle the hallucination phenomenon and the superiority of CIT-tuned VLMs over both CIEM and public datasets.

Abstract: Four-dimensional Digital Subtraction Angiography (4D DSA) plays a critical role in the diagnosis of many medical diseases, such as Arteriovenous Malformations (AVM) and Arteriovenous Fistulas (AVF). Despite its significant application value, the reconstruction of 4D DSA demands numerous views to effectively model the intricate vessels and radiocontrast flow, thereby implying a significant radiation dose. To address this high radiation issue, we propose a Time-aware Attenuation Voxel (TiAVox) approach for sparse-view 4D DSA reconstruction, which paves the way for high-quality 4D imaging. Additionally, 2D and 3D DSA imaging results can be generated from the reconstructed 4D DSA images. TiAVox introduces 4D attenuation voxel grids, which reflect attenuation properties from both spatial and temporal dimensions. It is optimized by minimizing discrepancies between the rendered images and sparse 2D DSA images. Without any neural network involved, TiAVox enjoys specific physical interpretability. The parameters of each learnable voxel represent the attenuation coefficients. We validated the TiAVox approach on both clinical and simulated datasets, achieving a 31.23 Peak Signal-to-Noise Ratio (PSNR) for novel view synthesis using only 30 views on the clinically sourced dataset, whereas traditional Feldkamp-Davis-Kress methods required 133 views. Similarly, with merely 10 views from the synthetic dataset, TiAVox yielded a PSNR of 34.32 for novel view synthesis and 41.40 for 3D reconstruction. We also executed ablation studies to corroborate the essential components of TiAVox. The code will be publically available.

Abstract: In this paper, we introduce a novel approach for generating texture images of infinite resolutions using Generative Adversarial Networks (GANs) based on a patch-by-patch paradigm. Existing texture synthesis techniques often rely on generating a large-scale texture using a one-forward pass to the generating model, this limits the scalability and flexibility of the generated images. In contrast, the proposed approach trains GANs models on a single texture image to generate relatively small patches that are locally correlated and can be seamlessly concatenated to form a larger image while using a constant GPU memory footprint. Our method learns the local texture structure and is able to generate arbitrary-size textures, while also maintaining coherence and diversity. The proposed method relies on local padding in the generator to ensure consistency between patches and utilizes spatial stochastic modulation to allow for local variations and diversity within the large-scale image. Experimental results demonstrate superior scalability compared to existing approaches while maintaining visual coherence of generated textures.

Abstract: We introduce the structured scene-text spotting task, which requires a scene-text OCR system to spot text in the wild according to a query regular expression. Contrary to generic scene text OCR, structured scene-text spotting seeks to dynamically condition both scene text detection and recognition on user-provided regular expressions. To tackle this task, we propose the Structured TExt sPotter (STEP), a model that exploits the provided text structure to guide the OCR process. STEP is able to deal with regular expressions that contain spaces and it is not bound to detection at the word-level granularity. Our approach enables accurate zero-shot structured text spotting in a wide variety of real-world reading scenarios and is solely trained on publicly available data. To demonstrate the effectiveness of our approach, we introduce a new challenging test dataset that contains several types of out-of-vocabulary structured text, reflecting important reading applications of fields such as prices, dates, serial numbers, license plates etc. We demonstrate that STEP can provide specialised OCR performance on demand in all tested scenarios.

Abstract: Dataset summarisation is a fruitful approach to dataset inspection. However, when applied to a single dataset the discovery of visual concepts is restricted to those most prominent. We argue that a comparative approach can expand upon this paradigm to enable richer forms of dataset inspection that go beyond the most prominent concepts. To enable dataset comparison we present a module that learns concept-level prototypes across datasets. We leverage self-supervised learning to discover these prototypes without supervision, and we demonstrate the benefits of our approach in two case-studies. Our findings show that dataset comparison extends dataset inspection and we hope to encourage more works in this direction. Code and usage instructions available at https://github.com/Nanne/ProtoSim

Abstract: Representing wild sounds as images is an important but challenging task due to the lack of paired datasets between sound and images and the significant differences in the characteristics of these two modalities. Previous studies have focused on generating images from sound in limited categories or music. In this paper, we propose a novel approach to generate images from in-the-wild sounds. First, we convert sound into text using audio captioning. Second, we propose audio attention and sentence attention to represent the rich characteristics of sound and visualize the sound. Lastly, we propose a direct sound optimization with CLIPscore and AudioCLIP and generate images with a diffusion-based model. In experiments, it shows that our model is able to generate high quality images from wild sounds and outperforms baselines in both quantitative and qualitative evaluations on wild audio datasets.

Abstract: We consider the visual disambiguation task of determining whether a pair of visually similar images depict the same or distinct 3D surfaces (e.g., the same or opposite sides of a symmetric building). Illusory image matches, where two images observe distinct but visually similar 3D surfaces, can be challenging for humans to differentiate, and can also lead 3D reconstruction algorithms to produce erroneous results. We propose a learning-based approach to visual disambiguation, formulating it as a binary classification task on image pairs. To that end, we introduce a new dataset for this problem, Doppelgangers, which includes image pairs of similar structures with ground truth labels. We also design a network architecture that takes the spatial distribution of local keypoints and matches as input, allowing for better reasoning about both local and global cues. Our evaluation shows that our method can distinguish illusory matches in difficult cases, and can be integrated into SfM pipelines to produce correct, disambiguated 3D reconstructions. See our project page for our code, datasets, and more results: http://doppelgangers-3d.github.io/.

Abstract: With the surge in attention to Egocentric Hand-Object Interaction (Ego-HOI), large-scale datasets such as Ego4D and EPIC-KITCHENS have been proposed. However, most current research is built on resources derived from third-person video action recognition. This inherent domain gap between first- and third-person action videos, which have not been adequately addressed before, makes current Ego-HOI suboptimal. This paper rethinks and proposes a new framework as an infrastructure to advance Ego-HOI recognition by Probing, Curation and Adaption (EgoPCA). We contribute comprehensive pre-train sets, balanced test sets and a new baseline, which are complete with a training-finetuning strategy. With our new framework, we not only achieve state-of-the-art performance on Ego-HOI benchmarks but also build several new and effective mechanisms and settings to advance further research. We believe our data and the findings will pave a new way for Ego-HOI understanding. Code and data are available at https://mvig-rhos.com/ego_pca

Abstract: Estimating the transferability of publicly available pretrained models to a target task has assumed an important place for transfer learning tasks in recent years. Existing efforts propose metrics that allow a user to choose one model from a pool of pre-trained models without having to fine-tune each model individually and identify one explicitly. With the growth in the number of available pre-trained models and the popularity of model ensembles, it also becomes essential to study the transferability of multiple-source models for a given target task. The few existing efforts study transferability in such multi-source ensemble settings using just the outputs of the classification layer and neglect possible domain or task mismatch. Moreover, they overlook the most important factor while selecting the source models, viz., the cohesiveness factor between them, which can impact the performance and confidence in the prediction of the ensemble. To address these gaps, we propose a novel Optimal tranSport-based suBmOdular tRaNsferability metric (OSBORN) to estimate the transferability of an ensemble of models to a downstream task. OSBORN collectively accounts for image domain difference, task difference, and cohesiveness of models in the ensemble to provide reliable estimates of transferability. We gauge the performance of OSBORN on both image classification and semantic segmentation tasks. Our setup includes 28 source datasets, 11 target datasets, 5 model architectures, and 2 pre-training methods. We benchmark our method against current state-of-the-art metrics MS-LEEP and E-LEEP, and outperform them consistently using the proposed approach.

Abstract: Recent years have witnessed great progress in creating vivid audio-driven portraits from monocular videos. However, how to seamlessly adapt the created video avatars to other scenarios with different backgrounds and lighting conditions remains unsolved. On the other hand, existing relighting studies mostly rely on dynamically lighted or multi-view data, which are too expensive for creating video portraits. To bridge this gap, we propose ReliTalk, a novel framework for relightable audio-driven talking portrait generation from monocular videos. Our key insight is to decompose the portrait's reflectance from implicitly learned audio-driven facial normals and images. Specifically, we involve 3D facial priors derived from audio features to predict delicate normal maps through implicit functions. These initially predicted normals then take a crucial part in reflectance decomposition by dynamically estimating the lighting condition of the given video. Moreover, the stereoscopic face representation is refined using the identity-consistent loss under simulated multiple lighting conditions, addressing the ill-posed problem caused by limited views available from a single monocular video. Extensive experiments validate the superiority of our proposed framework on both real and synthetic datasets. Our code is released in https://github.com/arthur-qiu/ReliTalk.

Abstract: Neural implicit representations have recently demonstrated compelling results on dense Simultaneous Localization And Mapping (SLAM) but suffer from the accumulation of errors in camera tracking and distortion in the reconstruction. Purposely, we present GO-SLAM, a deep-learning-based dense visual SLAM framework globally optimizing poses and 3D reconstruction in real-time. Robust pose estimation is at its core, supported by efficient loop closing and online full bundle adjustment, which optimize per frame by utilizing the learned global geometry of the complete history of input frames. Simultaneously, we update the implicit and continuous surface representation on-the-fly to ensure global consistency of 3D reconstruction. Results on various synthetic and real-world datasets demonstrate that GO-SLAM outperforms state-of-the-art approaches at tracking robustness and reconstruction accuracy. Furthermore, GO-SLAM is versatile and can run with monocular, stereo, and RGB-D input.