Computer Vision and Pattern Recognition (cs.CV)

Abstract: Language models demonstrate remarkable capacity to generalize representations learned in one modality to downstream tasks in other modalities. Can we trace this ability to individual neurons? We study the case where a frozen text transformer is augmented with vision using a self-supervised visual encoder and a single linear projection learned on an image-to-text task. Outputs of the projection layer are not immediately decodable into language describing image content; instead, we find that translation between modalities occurs deeper within the transformer. We introduce a procedure for identifying "multimodal neurons" that convert visual representations into corresponding text, and decoding the concepts they inject into the model's residual stream. In a series of experiments, we show that multimodal neurons operate on specific visual concepts across inputs, and have a systematic causal effect on image captioning.

Abstract: We generalize the class vectors found in neural networks to linear subspaces (i.e.~points in the Grassmann manifold) and show that the Grassmann Class Representation (GCR) enables the simultaneous improvement in accuracy and feature transferability. In GCR, each class is a subspace and the logit is defined as the norm of the projection of a feature onto the class subspace. We integrate Riemannian SGD into deep learning frameworks such that class subspaces in a Grassmannian are jointly optimized with the rest model parameters. Compared to the vector form, the representative capability of subspaces is more powerful. We show that on ImageNet-1K, the top-1 error of ResNet50-D, ResNeXt50, Swin-T and Deit3-S are reduced by 5.6%, 4.5%, 3.0% and 3.5%, respectively. Subspaces also provide freedom for features to vary and we observed that the intra-class feature variability grows when the subspace dimension increases. Consequently, we found the quality of GCR features is better for downstream tasks. For ResNet50-D, the average linear transfer accuracy across 6 datasets improves from 77.98% to 79.70% compared to the strong baseline of vanilla softmax. For Swin-T, it improves from 81.5% to 83.4% and for Deit3, it improves from 73.8% to 81.4%. With these encouraging results, we believe that more applications could benefit from the Grassmann class representation. Code is released at https://github.com/innerlee/GCR.

Abstract: In recent years, many deep learning-based methods have been proposed to tackle the problem of optical flow estimation and achieved promising results. However, they hardly consider that most videos are compressed and thus ignore the pre-computed information in compressed video streams. Motion vectors, one of the compression information, record the motion of the video frames. They can be directly extracted from the compression code stream without computational cost and serve as a solid prior for optical flow estimation. Therefore, we propose an optical flow model, MVFlow, which uses motion vectors to improve the speed and accuracy of optical flow estimation for compressed videos. In detail, MVFlow includes a key Motion-Vector Converting Module, which ensures that the motion vectors can be transformed into the same domain of optical flow and then be utilized fully by the flow estimation module. Meanwhile, we construct four optical flow datasets for compressed videos containing frames and motion vectors in pairs. The experimental results demonstrate the superiority of our proposed MVFlow, which can reduce the AEPE by 1.09 compared to existing models or save 52% time to achieve similar accuracy to existing models.

Abstract: Source-free domain adaptation (SFDA) aims to adapt a well-trained source model to an unlabelled target domain without accessing the source dataset, making it applicable in a variety of real-world scenarios. Existing SFDA methods ONLY assess their adapted models on the target training set, neglecting the data from unseen but identically distributed testing sets. This oversight leads to overfitting issues and constrains the model's generalization ability. In this paper, we propose a consistency regularization framework to develop a more generalizable SFDA method, which simultaneously boosts model performance on both target training and testing datasets. Our method leverages soft pseudo-labels generated from weakly augmented images to supervise strongly augmented images, facilitating the model training process and enhancing the generalization ability of the adapted model. To leverage more potentially useful supervision, we present a sampling-based pseudo-label selection strategy, taking samples with severer domain shift into consideration. Moreover, global-oriented calibration methods are introduced to exploit global class distribution and feature cluster information, further improving the adaptation process. Extensive experiments demonstrate our method achieves state-of-the-art performance on several SFDA benchmarks, and exhibits robustness on unseen testing datasets.

Abstract: Electron microscopy (EM) images exhibit anisotropic axial resolution due to the characteristics inherent to the imaging modality, presenting challenges in analysis and downstream tasks.In this paper, we propose a diffusion-model-based framework that overcomes the limitations of requiring reference data or prior knowledge about the degradation process. Our approach utilizes 2D diffusion models to consistently reconstruct 3D volumes and is well-suited for highly downsampled data. Extensive experiments conducted on two public datasets demonstrate the robustness and superiority of leveraging the generative prior compared to supervised learning methods. Additionally, we demonstrate our method's feasibility for self-supervised reconstruction, which can restore a single anisotropic volume without any training data.

Abstract: Herbal plants are nutritious plants that can be used as an alternative to traditional disease healing. In Indonesia there are various types of herbal plants. But with the development of the times, the existence of herbal plants as traditional medicines began to be forgotten so that not everyone could recognize them. Having the ability to identify herbal plants can have many positive impacts. However, there is a problem where identifying plants can take a long time because it requires in-depth knowledge and careful examination of plant criteria. So that the application of computer vision can help identify herbal plants. Previously, research had been conducted on the introduction of herbal plants from Vietnam using several algorithms, but from these research the accuracy was not high enough. Therefore, this study intends to implement transfer learning from the Convolutional Neural Network (CNN) algorithm to classify types of herbal plants from Indonesia. This research was conducted by collecting image data of herbal plants from Indonesia independently through the Google Images search engine. After that, it will go through the data preprocessing, classification using the transfer learning method from CNN, and analysis will be carried out. The CNN transfer learning models used are ResNet34, DenseNet121, and VGG11_bn. Based on the test results of the three models, it was found that DenseNet121 was the model with the highest accuracy, which was 87.4%. In addition, testing was also carried out using the scratch model and obtained an accuracy of 43.53%. The Hyperparameter configuration used in this test is the ExponentialLR scheduler with a gamma value of 0.9; learning rate 0.001; Cross Entropy Loss function; Adam optimizer; and the number of epochs is 50. Indonesia Medicinal Plant Dataset can be accessed at the following link https://github.com/Salmanim20/indo_medicinal_plant

Abstract: We present a pipeline for realistic embedding of virtual objects into footage of indoor scenes with focus on real-time AR applications. Our pipeline consists of two main components: A light estimator and a neural soft shadow texture generator. Our light estimation is based on deep neural nets and determines the main light direction, light color, ambient color and an opacity parameter for the shadow texture. Our neural soft shadow method encodes object-based realistic soft shadows as light direction dependent textures in a small MLP. We show that our pipeline can be used to integrate objects into AR scenes in a new level of realism in real-time. Our models are small enough to run on current mobile devices. We achieve runtimes of 9ms for light estimation and 5ms for neural shadows on an iPhone 11 Pro.

Abstract: Spatio-temporal action detection (STAD) aims to classify the actions present in a video and localize them in space and time. It has become a particularly active area of research in computer vision because of its explosively emerging real-world applications, such as autonomous driving, visual surveillance, entertainment, etc. Many efforts have been devoted in recent years to building a robust and effective framework for STAD. This paper provides a comprehensive review of the state-of-the-art deep learning-based methods for STAD. Firstly, a taxonomy is developed to organize these methods. Next, the linking algorithms, which aim to associate the frame- or clip-level detection results together to form action tubes, are reviewed. Then, the commonly used benchmark datasets and evaluation metrics are introduced, and the performance of state-of-the-art models is compared. At last, this paper is concluded, and a set of potential research directions of STAD are discussed.

Abstract: This paper introduces a new Convolutional Neural Network (ConvNet) architecture inspired by a class of partial differential equations (PDEs) called quasi-linear hyperbolic systems. With comparable performance on image classification task, it allows for the modification of the weights via a continuous group of symmetry. This is a significant shift from traditional models where the architecture and weights are essentially fixed. We wish to promote the (internal) symmetry as a new desirable property for a neural network, and to draw attention to the PDE perspective in analyzing and interpreting ConvNets in the broader Deep Learning community.

Abstract: In recent years, dominant Multi-object tracking (MOT) and segmentation (MOTS) methods mainly follow the tracking-by-detection paradigm. Transformer-based end-to-end (E2E) solutions bring some ideas to MOT and MOTS, but they cannot achieve a new state-of-the-art (SOTA) performance in major MOT and MOTS benchmarks. Detection and association are two main modules of the tracking-by-detection paradigm. Association techniques mainly depend on the combination of motion and appearance information. As deep learning has been recently developed, the performance of the detection and appearance model is rapidly improved. These trends made us consider whether we can achieve SOTA based on only high-performance detection and appearance model. Our paper mainly focuses on exploring this direction based on CBNetV2 with Swin-B as a detection model and MoCo-v2 as a self-supervised appearance model. Motion information and IoU mapping were removed during the association. Our method wins 1st place on the MOTS track and wins 2nd on the MOT track in the CVPR2023 WAD workshop. We hope our simple and effective method can give some insights to the MOT and MOTS research community. Source code will be released under this git repository

Abstract: An attractive book cover is important for the success of a book. In this paper, we apply Generative Adversarial Networks (GANs) to the book covers domain, using different methods for training in order to obtain better generated images. We interleave GANs with knowledge graphs to alter the input title to obtain multiple possible options for any given title, which are then used as an augmented input to the generator. Finally, we use the discriminator obtained during the training phase to select the best images generated with new titles. Our method performed better at generating book covers than previous attempts, and the knowledge graph gives better options to the book author or editor compared to using GANs alone.

Abstract: Recently, many breakthroughs are made in the field of Video Object Detection (VOD), but the performance is still limited due to the imaging limitations of RGB sensors in adverse illumination conditions. To alleviate this issue, this work introduces a new computer vision task called RGB-thermal (RGBT) VOD by introducing the thermal modality that is insensitive to adverse illumination conditions. To promote the research and development of RGBT VOD, we design a novel Erasure-based Interaction Network (EINet) and establish a comprehensive benchmark dataset (VT-VOD50) for this task. Traditional VOD methods often leverage temporal information by using many auxiliary frames, and thus have large computational burden. Considering that thermal images exhibit less noise than RGB ones, we develop a negative activation function that is used to erase the noise of RGB features with the help of thermal image features. Furthermore, with the benefits from thermal images, we rely only on a small temporal window to model the spatio-temporal information to greatly improve efficiency while maintaining detection accuracy. VT-VOD50 dataset consists of 50 pairs of challenging RGBT video sequences with complex backgrounds, various objects and different illuminations, which are collected in real traffic scenarios. Extensive experiments on VT-VOD50 dataset demonstrate the effectiveness and efficiency of our proposed method against existing mainstream VOD methods. The code of EINet and the dataset will be released to the public for free academic usage.

Abstract: Multi-view (or -modality) representation learning aims to understand the relationships between different view representations. Existing methods disentangle multi-view representations into consistent and view-specific representations by introducing strong inductive biases, which can limit their generalization ability. In this paper, we propose a novel multi-view representation disentangling method that aims to go beyond inductive biases, ensuring both interpretability and generalizability of the resulting representations. Our method is based on the observation that discovering multi-view consistency in advance can determine the disentangling information boundary, leading to a decoupled learning objective. We also found that the consistency can be easily extracted by maximizing the transformation invariance and clustering consistency between views. These observations drive us to propose a two-stage framework. In the first stage, we obtain multi-view consistency by training a consistent encoder to produce semantically-consistent representations across views as well as their corresponding pseudo-labels. In the second stage, we disentangle specificity from comprehensive representations by minimizing the upper bound of mutual information between consistent and comprehensive representations. Finally, we reconstruct the original data by concatenating pseudo-labels and view-specific representations. Our experiments on four multi-view datasets demonstrate that our proposed method outperforms 12 comparison methods in terms of clustering and classification performance. The visualization results also show that the extracted consistency and specificity are compact and interpretable. Our code can be found at \url{https://github.com/Guanzhou-Ke/DMRIB}.

Abstract: Electrocardiogram (ECG) is a widely used diagnostic tool for detecting heart conditions. Rare cardiac diseases may be underdiagnosed using traditional ECG analysis, considering that no training dataset can exhaust all possible cardiac disorders. This paper proposes using anomaly detection to identify any unhealthy status, with normal ECGs solely for training. However, detecting anomalies in ECG can be challenging due to significant inter-individual differences and anomalies present in both global rhythm and local morphology. To address this challenge, this paper introduces a novel multi-scale cross-restoration framework for ECG anomaly detection and localization that considers both local and global ECG characteristics. The proposed framework employs a two-branch autoencoder to facilitate multi-scale feature learning through a masking and restoration process, with one branch focusing on global features from the entire ECG and the other on local features from heartbeat-level details, mimicking the diagnostic process of cardiologists. Anomalies are identified by their high restoration errors. To evaluate the performance on a large number of individuals, this paper introduces a new challenging benchmark with signal point-level ground truths annotated by experienced cardiologists. The proposed method demonstrates state-of-the-art performance on this benchmark and two other well-known ECG datasets. The benchmark dataset and source code are available at: \url{https://github.com/MediaBrain-SJTU/ECGAD}

Abstract: Recent data-driven image colorization methods have enabled automatic or reference-based colorization, while still suffering from unsatisfactory and inaccurate object-level color control. To address these issues, we propose a new method called DiffColor that leverages the power of pre-trained diffusion models to recover vivid colors conditioned on a prompt text, without any additional inputs. DiffColor mainly contains two stages: colorization with generative color prior and in-context controllable colorization. Specifically, we first fine-tune a pre-trained text-to-image model to generate colorized images using a CLIP-based contrastive loss. Then we try to obtain an optimized text embedding aligning the colorized image and the text prompt, and a fine-tuned diffusion model enabling high-quality image reconstruction. Our method can produce vivid and diverse colors with a few iterations, and keep the structure and background intact while having colors well-aligned with the target language guidance. Moreover, our method allows for in-context colorization, i.e., producing different colorization results by modifying prompt texts without any fine-tuning, and can achieve object-level controllable colorization results. Extensive experiments and user studies demonstrate that DiffColor outperforms previous works in terms of visual quality, color fidelity, and diversity of colorization options.

Abstract: Using synthesized images to boost the performance of perception models is a long-standing research challenge in computer vision. It becomes more eminent in visual-centric autonomous driving systems with multi-view cameras as some long-tail scenarios can never be collected. Guided by the BEV segmentation layouts, the existing generative networks seem to synthesize photo-realistic street-view images when evaluated solely on scene-level metrics. However, once zoom-in, they usually fail to produce accurate foreground and background details such as heading. To this end, we propose a two-stage generative method, dubbed BEVControl, that can generate accurate foreground and background contents. In contrast to segmentation-like input, it also supports sketch style input, which is more flexible for humans to edit. In addition, we propose a comprehensive multi-level evaluation protocol to fairly compare the quality of the generated scene, foreground object, and background geometry. Our extensive experiments show that our BEVControl surpasses the state-of-the-art method, BEVGen, by a significant margin, from 5.89 to 26.80 on foreground segmentation mIoU. In addition, we show that using images generated by BEVControl to train the downstream perception model, it achieves on average 1.29 improvement in NDS score.

Abstract: 3D panoptic segmentation is a challenging perception task that requires both semantic segmentation and instance segmentation. In this task, we notice that images could provide rich texture, color, and discriminative information, which can complement LiDAR data for evident performance improvement, but their fusion remains a challenging problem. To this end, we propose LCPS, the first LiDAR-Camera Panoptic Segmentation network. In our approach, we conduct LiDAR-Camera fusion in three stages: 1) an Asynchronous Compensation Pixel Alignment (ACPA) module that calibrates the coordinate misalignment caused by asynchronous problems between sensors; 2) a Semantic-Aware Region Alignment (SARA) module that extends the one-to-one point-pixel mapping to one-to-many semantic relations; 3) a Point-to-Voxel feature Propagation (PVP) module that integrates both geometric and semantic fusion information for the entire point cloud. Our fusion strategy improves about 6.9% PQ performance over the LiDAR-only baseline on NuScenes dataset. Extensive quantitative and qualitative experiments further demonstrate the effectiveness of our novel framework. The code will be released at https://github.com/zhangzw12319/lcps.git.

Abstract: Class incremental learning (CIL) aims to incrementally update a trained model with the new classes of samples (plasticity) while retaining previously learned ability (stability). To address the most challenging issue in this goal, i.e., catastrophic forgetting, the mainstream paradigm is memory-replay CIL, which consolidates old knowledge by replaying a small number of old classes of samples saved in the memory. Despite effectiveness, the inherent destruction-reconstruction dynamics in memory-replay CIL are an intrinsic limitation: if the old knowledge is severely destructed, it will be quite hard to reconstruct the lossless counterpart. Our theoretical analysis shows that the destruction of old knowledge can be effectively alleviated by balancing the contribution of samples from the current phase and those saved in the memory. Motivated by this theoretical finding, we propose a novel Balanced Destruction-Reconstruction module (BDR) for memory-replay CIL, which can achieve better knowledge reconstruction by reducing the degree of maximal destruction of old knowledge. Specifically, to achieve a better balance between old knowledge and new classes, the proposed BDR module takes into account two factors: the variance in training status across different classes and the quantity imbalance of samples from the current phase and memory. By dynamically manipulating the gradient during training based on these factors, BDR can effectively alleviate knowledge destruction and improve knowledge reconstruction. Extensive experiments on a range of CIL benchmarks have shown that as a lightweight plug-and-play module, BDR can significantly improve the performance of existing state-of-the-art methods with good generalization.

Abstract: Feature selection could be defined as an optimization problem and solved by bio-inspired algorithms. Bees Algorithm (BA) shows decent performance in feature selection optimization tasks. On the other hand, Local Phase Quantization (LPQ) is a frequency domain feature which has excellent performance on Depth images. Here, after extracting LPQ features out of RGB (colour) and Depth images from the Iranian Kinect Face Database (IKFDB), the Bees feature selection algorithm applies to select the desired number of features for final classification tasks. IKFDB is recorded with Kinect sensor V.2 and contains colour and depth images for facial and facial micro-expressions recognition purposes. Here five facial expressions of Anger, Joy, Surprise, Disgust and Fear are used for final validation. The proposed Bees LPQ method is compared with Particle Swarm Optimization (PSO) LPQ, PCA LPQ, Lasso LPQ, and just LPQ features for classification tasks with Support Vector Machines (SVM), K-Nearest Neighbourhood (KNN), Shallow Neural Network and Ensemble Subspace KNN. Returned results, show a decent performance of the proposed algorithm (99 % accuracy) in comparison with others.

Abstract: 3D semantic scene understanding tasks have achieved great success with the emergence of deep learning, but often require a huge amount of manually annotated training data. To alleviate the annotation cost, we propose the first weakly-supervised 3D instance segmentation method that only requires categorical semantic labels as supervision, and we do not need instance-level labels. The required semantic annotations can be either dense or extreme sparse (e.g. 0.02% of total points). Even without having any instance-related ground-truth, we design an approach to break point clouds into raw fragments and find the most confident samples for learning instance centroids. Furthermore, we construct a recomposed dataset using pseudo instances, which is used to learn our defined multilevel shape-aware objectness signal. An asymmetrical object inference algorithm is followed to process core points and boundary points with different strategies, and generate high-quality pseudo instance labels to guide iterative training. Experiments demonstrate that our method can achieve comparable results with recent fully supervised methods. By generating pseudo instance labels from categorical semantic labels, our designed approach can also assist existing methods for learning 3D instance segmentation at reduced annotation cost.

Abstract: Visibility in hazy nighttime scenes is frequently reduced by multiple factors, including low light, intense glow, light scattering, and the presence of multicolored light sources. Existing nighttime dehazing methods often struggle with handling glow or low-light conditions, resulting in either excessively dark visuals or unsuppressed glow outputs. In this paper, we enhance the visibility from a single nighttime haze image by suppressing glow and enhancing low-light regions. To handle glow effects, our framework learns from the rendered glow pairs. Specifically, a light source aware network is proposed to detect light sources of night images, followed by the APSF (Angular Point Spread Function)-guided glow rendering. Our framework is then trained on the rendered images, resulting in glow suppression. Moreover, we utilize gradient-adaptive convolution, to capture edges and textures in hazy scenes. By leveraging extracted edges and textures, we enhance the contrast of the scene without losing important structural details. To boost low-light intensity, our network learns an attention map, then adjusted by gamma correction. This attention has high values on low-light regions and low values on haze and glow regions. Extensive evaluation on real nighttime haze images, demonstrates the effectiveness of our method. Our experiments demonstrate that our method achieves a PSNR of 30.72dB, outperforming state-of-the-art methods by 14$\%$ on GTA5 nighttime haze dataset. Our data and code is available at: \url{https://github.com/jinyeying/nighttime_dehaze}.

Abstract: We introduce PoissonNet, an architecture for shape reconstruction that addresses the challenge of recovering 3D shapes from points. Traditional deep neural networks face challenges with common 3D shape discretization techniques due to their computational complexity at higher resolutions. To overcome this, we leverage Fourier Neural Operators (FNOs) to solve the Poisson equation and reconstruct a mesh from oriented point cloud measurements. PoissonNet exhibits two main advantages. First, it enables efficient training on low-resolution data while achieving comparable performance at high-resolution evaluation, thanks to the resolution-agnostic nature of FNOs. This feature allows for one-shot super-resolution. Second, our method surpasses existing approaches in reconstruction quality while being differentiable. Overall, our proposed method not only improves upon the limitations of classical deep neural networks in shape reconstruction but also achieves superior results in terms of reconstruction quality, running time, and resolution flexibility. Furthermore, we demonstrate that the Poisson surface reconstruction problem is well-posed in the limit case by showing a universal approximation theorem for the solution operator of the Poisson equation with distributional data utilizing the Fourier Neuronal Operator, which provides a theoretical foundation for our numerical results. The code to reproduce the experiments is available on: \url{https://github.com/arsenal9971/PoissonNet}.

Abstract: Tensor decompositions are powerful tools for analyzing multi-dimensional data in their original format. Besides tensor decompositions like Tucker and CP, Tensor SVD (t-SVD) which is based on the t-product of tensors is another extension of SVD to tensors that recently developed and has found numerous applications in analyzing high dimensional data. This paper offers a new insight into the t-Product and shows that this product is a block convolution of two tensors with periodic boundary conditions. Based on this viewpoint, we propose a new tensor-tensor product called the $\star_c{}\text{-Product}$ based on Block convolution with reflective boundary conditions. Using a tensor framework, this product can be easily extended to tensors of arbitrary order. Additionally, we introduce a tensor decomposition based on our $\star_c{}\text{-Product}$ for arbitrary order tensors. Compared to t-SVD, our new decomposition has lower complexity, and experiments show that it yields higher-quality results in applications such as classification and compression.

Abstract: Weakly-supervised image segmentation has recently attracted increasing research attentions, aiming to avoid the expensive pixel-wise labeling. In this paper, we present an effective method, namely Point2Mask, to achieve high-quality panoptic prediction using only a single random point annotation per target for training. Specifically, we formulate the panoptic pseudo-mask generation as an Optimal Transport (OT) problem, where each ground-truth (gt) point label and pixel sample are defined as the label supplier and consumer, respectively. The transportation cost is calculated by the introduced task-oriented maps, which focus on the category-wise and instance-wise differences among the various thing and stuff targets. Furthermore, a centroid-based scheme is proposed to set the accurate unit number for each gt point supplier. Hence, the pseudo-mask generation is converted into finding the optimal transport plan at a globally minimal transportation cost, which can be solved via the Sinkhorn-Knopp Iteration. Experimental results on Pascal VOC and COCO demonstrate the promising performance of our proposed Point2Mask approach to point-supervised panoptic segmentation. Source code is available at: https://github.com/LiWentomng/Point2Mask.

Abstract: Cooperative perception can effectively enhance individual perception performance by providing additional viewpoint and expanding the sensing field. Existing cooperation paradigms are either interpretable (result cooperation) or flexible (feature cooperation). In this paper, we propose the concept of query cooperation to enable interpretable instance-level flexible feature interaction. To specifically explain the concept, we propose a cooperative perception framework, termed QUEST, which let query stream flow among agents. The cross-agent queries are interacted via fusion for co-aware instances and complementation for individual unaware instances. Taking camera-based vehicle-infrastructure perception as a typical practical application scene, the experimental results on the real-world dataset, DAIR-V2X-Seq, demonstrate the effectiveness of QUEST and further reveal the advantage of the query cooperation paradigm on transmission flexibility and robustness to packet dropout. We hope our work can further facilitate the cross-agent representation interaction for better cooperative perception in practice.

Abstract: Dietary assessment is a key contributor to monitoring health status. Existing self-report methods are tedious and time-consuming with substantial biases and errors. Image-based food portion estimation aims to estimate food energy values directly from food images, showing great potential for automated dietary assessment solutions. Existing image-based methods either use a single-view image or incorporate multi-view images and depth information to estimate the food energy, which either has limited performance or creates user burdens. In this paper, we propose an end-to-end deep learning framework for food energy estimation from a monocular image through 3D shape reconstruction. We leverage a generative model to reconstruct the voxel representation of the food object from the input image to recover the missing 3D information. Our method is evaluated on a publicly available food image dataset Nutrition5k, resulting a Mean Absolute Error (MAE) of 40.05 kCal and Mean Absolute Percentage Error (MAPE) of 11.47% for food energy estimation. Our method uses RGB image as the only input at the inference stage and achieves competitive results compared to the existing method requiring both RGB and depth information.

Abstract: Fine-grained image classification (FGIC) is a challenging task in computer vision for due to small visual differences among inter-subcategories, but, large intra-class variations. Deep learning methods have achieved remarkable success in solving FGIC. In this paper, we propose a fusion approach to address FGIC by combining global texture with local patch-based information. The first pipeline extracts deep features from various fixed-size non-overlapping patches and encodes features by sequential modelling using the long short-term memory (LSTM). Another path computes image-level textures at multiple scales using the local binary patterns (LBP). The advantages of both streams are integrated to represent an efficient feature vector for image classification. The method is tested on eight datasets representing the human faces, skin lesions, food dishes, marine lives, etc. using four standard backbone CNNs. Our method has attained better classification accuracy over existing methods with notable margins.

Abstract: Text-to-motion generation has gained increasing attention, but most existing methods are limited to generating short-term motions that correspond to a single sentence describing a single action. However, when a text stream describes a sequence of continuous motions, the generated motions corresponding to each sentence may not be coherently linked. Existing long-term motion generation methods face two main issues. Firstly, they cannot directly generate coherent motions and require additional operations such as interpolation to process the generated actions. Secondly, they generate subsequent actions in an autoregressive manner without considering the influence of future actions on previous ones. To address these issues, we propose a novel approach that utilizes a past-conditioned diffusion model with two optional coherent sampling methods: Past Inpainting Sampling and Compositional Transition Sampling. Past Inpainting Sampling completes subsequent motions by treating previous motions as conditions, while Compositional Transition Sampling models the distribution of the transition as the composition of two adjacent motions guided by different text prompts. Our experimental results demonstrate that our proposed method is capable of generating compositional and coherent long-term 3D human motions controlled by a user-instructed long text stream. The code is available at \href{https://github.com/yangzhao1230/PCMDM}{https://github.com/yangzhao1230/PCMDM}.

Abstract: Understanding 3d human interactions is fundamental for fine-grained scene analysis and behavioural modeling. However, most of the existing models predict incorrect, lifeless 3d estimates, that miss the subtle human contact aspects--the essence of the event--and are of little use for detailed behavioral understanding. This paper addresses such issues with several contributions: (1) we introduce models for interaction signature estimation (ISP) encompassing contact detection, segmentation, and 3d contact signature prediction; (2) we show how such components can be leveraged to ensure contact consistency during 3d reconstruction; (3) we construct several large datasets for learning and evaluating 3d contact prediction and reconstruction methods; specifically, we introduce CHI3D, a lab-based accurate 3d motion capture dataset with 631 sequences containing $2,525$ contact events, $728,664$ ground truth 3d poses, as well as FlickrCI3D, a dataset of $11,216$ images, with $14,081$ processed pairs of people, and $81,233$ facet-level surface correspondences. Finally, (4) we propose methodology for recovering the ground-truth pose and shape of interacting people in a controlled setup and (5) annotate all 3d interaction motions in CHI3D with textual descriptions. Motion data in multiple formats (GHUM and SMPLX parameters, Human3.6m 3d joints) is made available for research purposes at \url{https://ci3d.imar.ro}, together with an evaluation server and a public benchmark.

Abstract: Object detection is a vital task in computer vision and has become an integral component of numerous critical systems. However, state-of-the-art object detectors, similar to their classification counterparts, are susceptible to small adversarial perturbations that can significantly alter their normal behavior. Unlike classification, the robustness of object detectors has not been thoroughly explored. In this work, we take the initial step towards bridging the gap between the robustness of classification and object detection by leveraging adversarially trained classification models. Merely utilizing adversarially trained models as backbones for object detection does not result in robustness. We propose effective modifications to the classification-based backbone to instill robustness in object detection without incurring any computational overhead. To further enhance the robustness achieved by the proposed modified backbone, we introduce two lightweight components: imitation loss and delayed adversarial training. Extensive experiments on the MS-COCO and Pascal VOC datasets are conducted to demonstrate the effectiveness of our proposed approach.

Abstract: Multi-label image recognition in the low-label regime is a task of great challenge and practical significance. Previous works have focused on learning the alignment between textual and visual spaces to compensate for limited image labels, yet may suffer from reduced accuracy due to the scarcity of high-quality multi-label annotations. In this research, we leverage the powerful alignment between textual and visual features pretrained with millions of auxiliary image-text pairs. We introduce an efficient and effective framework called Evidence-guided Dual Context Optimization (DualCoOp++), which serves as a unified approach for addressing partial-label and zero-shot multi-label recognition. In DualCoOp++ we separately encode evidential, positive, and negative contexts for target classes as parametric components of the linguistic input (i.e., prompts). The evidential context aims to discover all the related visual content for the target class, and serves as guidance to aggregate positive and negative contexts from the spatial domain of the image, enabling better distinguishment between similar categories. Additionally, we introduce a Winner-Take-All module that promotes inter-class interaction during training, while avoiding the need for extra parameters and costs. As DualCoOp++ imposes minimal additional learnable overhead on the pretrained vision-language framework, it enables rapid adaptation to multi-label recognition tasks with limited annotations and even unseen classes. Experiments on standard multi-label recognition benchmarks across two challenging low-label settings demonstrate the superior performance of our approach compared to state-of-the-art methods.

Abstract: Rigorously testing autonomy systems is essential for making safe self-driving vehicles (SDV) a reality. It requires one to generate safety critical scenarios beyond what can be collected safely in the world, as many scenarios happen rarely on public roads. To accurately evaluate performance, we need to test the SDV on these scenarios in closed-loop, where the SDV and other actors interact with each other at each timestep. Previously recorded driving logs provide a rich resource to build these new scenarios from, but for closed loop evaluation, we need to modify the sensor data based on the new scene configuration and the SDV's decisions, as actors might be added or removed and the trajectories of existing actors and the SDV will differ from the original log. In this paper, we present UniSim, a neural sensor simulator that takes a single recorded log captured by a sensor-equipped vehicle and converts it into a realistic closed-loop multi-sensor simulation. UniSim builds neural feature grids to reconstruct both the static background and dynamic actors in the scene, and composites them together to simulate LiDAR and camera data at new viewpoints, with actors added or removed and at new placements. To better handle extrapolated views, we incorporate learnable priors for dynamic objects, and leverage a convolutional network to complete unseen regions. Our experiments show UniSim can simulate realistic sensor data with small domain gap on downstream tasks. With UniSim, we demonstrate closed-loop evaluation of an autonomy system on safety-critical scenarios as if it were in the real world.

Abstract: This paper presents an improved DETR detector that maintains a "plain" nature: using a single-scale feature map and global cross-attention calculations without specific locality constraints, in contrast to previous leading DETR-based detectors that reintroduce architectural inductive biases of multi-scale and locality into the decoder. We show that two simple technologies are surprisingly effective within a plain design to compensate for the lack of multi-scale feature maps and locality constraints. The first is a box-to-pixel relative position bias (BoxRPB) term added to the cross-attention formulation, which well guides each query to attend to the corresponding object region while also providing encoding flexibility. The second is masked image modeling (MIM)-based backbone pre-training which helps learn representation with fine-grained localization ability and proves crucial for remedying dependencies on the multi-scale feature maps. By incorporating these technologies and recent advancements in training and problem formation, the improved "plain" DETR showed exceptional improvements over the original DETR detector. By leveraging the Object365 dataset for pre-training, it achieved 63.9 mAP accuracy using a Swin-L backbone, which is highly competitive with state-of-the-art detectors which all heavily rely on multi-scale feature maps and region-based feature extraction. Code is available at https://github.com/impiga/Plain-DETR .

Abstract: Depth completion, which aims to generate high-quality dense depth maps from sparse depth maps, has attracted increasing attention in recent years. Previous work usually employs RGB images as guidance, and introduces iterative spatial propagation to refine estimated coarse depth maps. However, most of the propagation refinement methods require several iterations and suffer from a fixed receptive field, which may contain irrelevant and useless information with very sparse input. In this paper, we address these two challenges simultaneously by revisiting the idea of deformable convolution. We propose an effective architecture that leverages deformable kernel convolution as a single-pass refinement module, and empirically demonstrate its superiority. To better understand the function of deformable convolution and exploit it for depth completion, we further systematically investigate a variety of representative strategies. Our study reveals that, different from prior work, deformable convolution needs to be applied on an estimated depth map with a relatively high density for better performance. We evaluate our model on the large-scale KITTI dataset and achieve state-of-the-art level performance in both accuracy and inference speed. Our code is available at https://github.com/AlexSunNik/ReDC.

Abstract: We present the All-Seeing (AS) project: a large-scale data and model for recognizing and understanding everything in the open world. Using a scalable data engine that incorporates human feedback and efficient models in the loop, we create a new dataset (AS-1B) with over 1 billion regions annotated with semantic tags, question-answering pairs, and detailed captions. It covers a wide range of 3.5 million common and rare concepts in the real world, and has 132.2 billion tokens that describe the concepts and their attributes. Leveraging this new dataset, we develop the All-Seeing model (ASM), a unified framework for panoptic visual recognition and understanding. The model is trained with open-ended language prompts and locations, which allows it to generalize to various vision and language tasks with remarkable zero-shot performance, including region-text retrieval, region recognition, captioning, and question-answering. We hope that this project can serve as a foundation for vision-language artificial general intelligence research. Models and the dataset shall be released at https://github.com/OpenGVLab/All-Seeing, and demo can be seen at https://huggingface.co/spaces/OpenGVLab/all-seeing.