ICCV 2023 Papers

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 annotate 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 automatic, 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 highlights 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.

Multi-modality image fusion and segmentation play a vital role in autonomous driving and robotic operation. Early efforts focus on boosting the performance for only one task, e.g., fusion or segmentation, making it hard to reach `Best of Both Worlds'. To overcome this issue, in this paper, we propose a Multi-interactive Feature learning architecture for image fusion and segmentation, namely SegMiF, and exploit dual-task correlation to promote the performance of both tasks. The SegMiF is of a cascade structure, containing a fusion sub-network and a commonly used segmentation sub-network. By slickly bridging intermediate features between two components, the knowledge learned from the segmentation task can effectively assist the fusion task. Also, the benefited fusion network supports the segmentation one to perform more pretentiously. Besides, a hierarchical interactive attention block is established to ensure fine-grained mapping of all the vital information between two tasks, so that the modality/semantic features can be fully mutual-interactive. In addition, a dynamic weight factor is introduced to automatically adjust the corresponding weights of each task, which can balance the interactive feature correspondence and break through the limitation of laborious tuning. Furthermore, we construct a smart multi-wave binocular imaging system and collect a full-time multi-modality benchmark with 15 annotated pixel-level categories for image fusion and segmentation. Extensive experiments on several public datasets and our benchmark demonstrate that the proposed method outputs visually appealing fused images and perform averagely 7.66% higher segmentation mIoU in the real-world scene than the state-of-the-art approaches. The source code and benchmark are available at https://github.com/JinyuanLiu-CV/SegMiF.

Interactive segmentation enables users to segment as needed by providing cues of objects, which introduces human-computer interaction for many fields, such as image editing and medical image analysis. Typically, massive and expansive pixel-level annotations are spent to train deep models by object-oriented interactions with manually labeled object masks. In this work, we reveal that informative interactions can be made by simulation with semantic-consistent yet diverse region exploration in an unsupervised paradigm. Concretely, we introduce a Multi-granularity Interaction Simulation (MIS) approach to open up a promising direction for unsupervised interactive segmentation. Drawing on the high-quality dense features produced by recent self-supervised models, we propose to gradually merge patches or regions with similar features to form more extensive regions and thus, every merged region serves as a semantic-meaningful multi-granularity proposal. By randomly sampling these proposals and simulating possible interactions based on them, we provide meaningful interaction at multiple granularities to teach the model to understand interactions. Our MIS significantly outperforms non-deep learning unsupervised methods and is even comparable with some previous deep-supervised methods without any annotation.

Few-Shot Video Object Segmentation (FSVOS) aims to segment objects in a query video with the same category defined by a few annotated support images. However, this task was seldom explored. In this work, based on IPMT, a state-of-the-art few-shot image segmentation method that combines external support guidance information with adaptive query guidance cues, we propose to leverage multi-grained temporal guidance information for handling the temporal correlation nature of video data. We decompose the query video information into a clip prototype and a memory prototype for capturing local and long-term internal temporal guidance, respectively. Frame prototypes are further used for each frame independently to handle fine-grained adaptive guidance and enable bidirectional clip-frame prototype communication. To reduce the influence of noisy memory, we propose to leverage the structural similarity relation among different predicted regions and the support for selecting reliable memory frames. Furthermore, a new segmentation loss is also proposed to enhance the category discriminability of the learned prototypes. Experimental results demonstrate that our proposed video IPMT model significantly outperforms previous models on two benchmark datasets. Code is available at https://github.com/nankepan/VIPMT.

Traditional and deep Structure-from-Motion (SfM) methods typically operate under the assumption that the scene is rigid, i.e., the environment is static or consists of a single moving object. Few multi-body SfM approaches address the reconstruction of multiple rigid bodies in a scene but suffer from the inherent scale ambiguity of SfM, such that objects are reconstructed at inconsistent scales. We propose a depth and camera pose estimation framework to resolve the scale ambiguity in multi-body scenes. Specifically, starting from disorganized images, we present a novel multi-view scale estimator that resolves the camera pose ambiguity and a multi-body plane sweep network that generalizes depth estimation to dynamic scenes. Experiments demonstrate the advantages of our method over state-of-the-art SfM frameworks in multi-body scenes and show that it achieves comparable results in static scenes. The code and dataset are available at https://github.com/andreadalcin/MultiBodySfM.

Despite the remarkable progress of Fine-grained visual classification (FGVC) with years of history, it is still limited to recognizing 2 images. Recognizing objects in the physical world (i.e., 3D environment) poses a unique challenge -- discriminative information is not only present in visible local regions but also in other unseen views. Therefore, in addition to finding the distinguishable part from the current view, efficient and accurate recognition requires inferring the critical perspective with minimal glances. E.g., a person might recognize a "Ford sedan" with a glance at its side and then know that looking at the front can help tell which model it is. In this paper, towards FGVC in the real physical world, we put forward the problem of multi-view active fine-grained visual recognition (MAFR) and complete this study in three steps: (i) a multi-view, fine-grained vehicle dataset is collected as the testbed, (ii) a pilot experiment is designed to validate the need and research value of MAFR, (iii) a policy-gradient-based framework along with a dynamic exiting strategy is proposed to achieve efficient recognition with active view selection. Our comprehensive experiments demonstrate that the proposed method outperforms previous multi-view recognition works and can extend existing state-of-the-art FGVC methods and advanced neural networks to become FGVC experts in the 3D environment.

While multi-task learning (MTL) has become an attractive topic, its training usually poses more difficulties than the single-task case. How to successfully apply knowledge distillation into MTL to improve training efficiency and model performance is still a challenging problem. In this paper, we introduce a new knowledge distillation procedure with an alternative match for MTL of dense prediction based on two simple design principles. First, for memory and training efficiency, we use a single strong multi-task model as a teacher during training instead of multiple teachers, as widely adopted in existing studies. Second, we employ a less sensitive Cauchy-Schwarz (CS) divergence instead of the Kullback-Leibler (KL) divergence and propose a CS distillation loss accordingly. With the less sensitive divergence, our knowledge distillation with an alternative match is applied for capturing inter-task and intra-task information between the teacher model and the student model of each task, thereby learning more "dark knowledge" for effective distillation. We conducted extensive experiments on dense prediction datasets, including NYUD-v2 and PASCAL-Context, for multiple vision tasks, such as semantic segmentation, human parts segmentation, depth estimation, surface normal estimation, and boundary detection. The results show that our proposed method decidedly improves model performance and the practical inference efficiency.

We present a simple and effective Multi-scale Residual Low-Pass Filter Network (MRLPFNet) that jointly explores the image details and main structures for image deblurring. Our work is motivated by an observation that the difference between the blurry image and the clear one not only contains high-frequency contents(Note that the high-frequency contents in an image correspond to the image details, while the low-frequency ones denote the main structures of an image.) but also includes low-frequency information due to the influence of blur, while using the standard residual learning is less effective for modeling the main structure distorted by the blur. Considering that the low-frequency contents usually correspond to main global structures that are spatially variant, we first propose a learnable low-pass filter based on a self-attention mechanism to adaptively explore the global contexts for better modeling the low-frequency information. Then we embed it into a Residual Low-Pass Filter (RLPF) module, which involves an additional fully convolutional neural network with the standard residual learning to model the high-frequency information. We formulate the RLPF module into an end-to-end trainable network based on an encoder and decoder architecture and develop a wavelet-based feature fusion to fuse the multi-scale features. Experimental results show that our method performs favorably against state-of-the-art ones on commonly-used benchmarks.

Scene flow estimation provides the fundamental motion perception of a dynamic scene, which is of practical importance in many computer vision applications. In this paper, we propose a novel multi-scale bidirectional recurrent architecture that iteratively optimizes the coarse-to-fine scene flow estimation. In each resolution scale of estimation, a novel bidirectional gated recurrent unit is proposed to bidirectionally and iteratively augment point features and produce progressively optimized scene flow. The optimization of each iteration is integrated with the hybrid correlation that captures not only local correlation but also semantic correlation for more accurate estimation. Experimental results indicate that our proposed architecture significantly outperforms the existing state-of-the-art approaches on both FlyingThings3D and KITTI benchmarks while maintaining superior time efficiency. Codes and pre-trained models are publicly available at https://github.com/cwc1260/MSBRN.

Understanding and mapping a new environment are core abilities of any autonomously navigating agent. While classical robotics usually estimates maps in a stand-alone manner with SLAM variants, which maintain a topological or metric representation, end-to-end learning of navigation keeps some form of memory in a neural network. Networks are typically imbued with inductive biases, which can range from vectorial representations to birds-eye metric tensors or topological structures. In this work, we propose to structure neural networks with two neural implicit representations, which are learned dynamically during each episode and map the content of the scene: (i) the Semantic Finder predicts the position of a previously seen queried object; (ii) the Occupancy and Exploration Implicit Representation encapsulates information about explored area and obstacles, and is queried with a novel global read mechanism which directly maps from function space to a usable embedding space. Both representations are leveraged by an agent trained with Reinforcement Learning (RL) and learned online during each episode. We evaluate the agent on Multi-Object Navigation and show the high impact of using neural implicit representations as a memory source.

Recent work in unsupervised multi-object segmentation shows impressive results by predicting motion from a single image despite the inherent ambiguity in predicting motion without the next image. On the other hand, the set of possible motions for an image can be constrained to a low-dimensional space by considering the scene structure and moving objects in it. We propose to model pixel-wise geometry and object motion to remove ambiguity in reconstructing flow from a single image. Specifically, we divide the image into coherently moving regions and use depth to construct flow bases that best explain the observed flow in each region. We achieve state-of-the-art results in unsupervised multi-object segmentation on synthetic and real-world datasets by modeling the scene structure and object motion. Our evaluation of the predicted depth maps shows reliable performance in monocular depth estimation.

Light-weight time-of-flight (ToF) depth sensors are compact and cost-efficient, and thus widely used on mobile devices for tasks such as autofocus and obstacle detection. However, due to the sparse and noisy depth measurements, these sensors have rarely been considered for dense geometry reconstruction. In this work, we present the first dense SLAM system with a monocular camera and a light-weight ToF sensor. Specifically, we propose a multi-modal implicit scene representation that supports rendering both the signals from the RGB camera and light-weight ToF sensor which drives the optimization by comparing with the raw sensor inputs. Moreover, in order to guarantee successful pose tracking and reconstruction, we exploit a predicted depth as an intermediate supervision and develop a coarse-to-fine optimization strategy for efficient learning of the implicit representation. At last, the temporal information is explicitly exploited to deal with the noisy signals from light-weight ToF sensors to improve the accuracy and robustness of the system. Experiments demonstrate that our system well exploits the signals of light-weight ToF sensors and achieves competitive results both on camera tracking and dense scene reconstruction. Project page: https://zju3dv.github.io/tof_slam/.

Infrared and visible image fusion aims to integrate comprehensive information from multiple sources to achieve superior performances on various practical tasks, such as detection, over that of a single modality. However, most existing methods directly combined the texture details and object contrast of different modalities, ignoring the dynamic changes in reality, which diminishes the visible texture in good lighting conditions and the infrared contrast in low lighting conditions. To fill this gap, we propose a dynamic image fusion framework with a multi-modal gated mixture of local-to-global experts, termed MoE-Fusion, to dynamically extract effective and comprehensive information from the respective modalities. Our model consists of a Mixture of Local Experts (MoLE) and a Mixture of Global Experts (MoGE) guided by a multi-modal gate. The MoLE performs specialized learning of multi-modal local features, prompting the fused images to retain the local information in a sample-adaptive manner, while the MoGE focuses on the global information that complements the fused image with overall texture detail and contrast. Extensive experiments show that our MoE-Fusion outperforms state-of-the-art methods in preserving multi-modal image texture and contrast through the local-to-global dynamic learning paradigm, and also achieves superior performance on detection tasks. Our code is available: https://github.com/SunYM2020/MoE-Fusion.

Continual Test-Time Adaptation (CTTA) generalizes conventional Test-Time Adaptation (TTA) by assuming that the target domain is dynamic over time rather than stationary. In this paper, we explore Multi-Modal Continual Test-Time Adaptation (MM-CTTA) as a new extension of CTTA for 3D semantic segmentation. The key to MM-CTTA is to adaptively attend to the reliable modality while avoiding catastrophic forgetting during continual domain shifts, which is out of the capability of previous TTA or CTTA methods. To fulfill this gap, we propose an MM-CTTA method called Continual Cross-Modal Adaptive Clustering (CoMAC) that addresses this task from two perspectives. On one hand, we propose an adaptive dual-stage mechanism to generate reliable cross-modal predictions by attending to the reliable modality based on the class-wise feature-centroid distance in the latent space. On the other hand, to perform test-time adaptation without catastrophic forgetting, we design class-wise momentum queues that capture confident target features for adaptation while stochastically restoring pseudo-source features to revisit source knowledge. We further introduce two new benchmarks to facilitate the exploration of MM-CTTA in the future. Our experimental results show that our method achieves state-of-the-art performance on both benchmarks. Visit our project website at https://sites.google.com/view/mmcotta.

The decentralized and privacy-preserving nature of federated learning (FL) makes it vulnerable to backdoor attacks aiming to manipulate the behavior of the resulting model on specific adversary-chosen inputs. However, most existing defenses based on statistical differences take effect only against specific attacks, especially when the malicious gradients are similar to benign ones or the data are highly non-independent and identically distributed (non-IID). In this paper, we revisit the distance-based defense methods and discover that i) Euclidean distance becomes meaningless in high dimensions and ii) malicious gradients with diverse characteristics cannot be identified by a single metric. To this end, we present a simple yet effective defense strategy with multi-metrics and dynamic weighting to identify backdoors adaptively. Furthermore, our novel defense has no reliance on predefined assumptions over attack settings or data distributions and little impact on benign performance. To evaluate the effectiveness of our approach, we conduct comprehensive experiments on different datasets under various attack settings, where our method achieves the best defensive performance. For instance, we achieve the lowest backdoor accuracy of 3.06% under the difficult Edge-case PGD, showing significant superiority over previous defenses. The results also demonstrate that our method can be well-adapted to a wide range of non-IID degrees without sacrificing the benign performance.

Self-supervised learning (SSL) methods targeting scene images have seen a rapid growth recently, and they mostly rely on either a dedicated dense matching mechanism or a costly unsupervised object discovery module. This paper shows that instead of hinging on these strenuous operations, quality image representations can be learned by treating scene/multi-label image SSL simply as a multi-label classification problem, which greatly simplifies the learning framework. Specifically, multiple binary pseudo-labels are assigned for each input image by comparing its embeddings with those in two dictionaries, and the network is optimized using the binary cross entropy loss. The proposed method is named Multi-Label Self-supervised learning (MLS). Visualizations qualitatively show that clearly the pseudo-labels by MLS can automatically find semantically similar pseudo-positive pairs across different images to facilitate contrastive learning. MLS learns high quality representations on MS-COCO and achieves state-of-the-art results on classification, detection and segmentation benchmarks. At the same time, MLS is much simpler than existing methods, making it easier to deploy and for further exploration.

Existing knowledge distillation methods typically work by imparting the knowledge of output logits or intermediate feature maps from the teacher network to the student network, which is very successful in multi-class single-label learning. However, these methods can hardly be extended to the multi-label learning scenario, where each instance is associated with multiple semantic labels, because the prediction probabilities do not sum to one and feature maps of the whole example may ignore minor classes in such a scenario. In this paper, we propose a novel multi-label knowledge distillation method. On one hand, it exploits the informative semantic knowledge from the logits by dividing the multi-label learning problem into a set of binary classification problems; on the other hand, it enhances the distinctiveness of the learned feature representations by leveraging the structural information of label-wise embeddings. Experimental results on multiple benchmark datasets validate that the proposed method can avoid knowledge counteraction among labels, thus achieving superior performance against diverse comparing methods.

CNN's limited receptive field restricts its ability to capture long-range spatial-temporal dependencies, leading to unsatisfactory performance in video super-resolution. To tackle this challenge, this paper presents a novel multi-frequency representation enhancement module (MFE) that performs spatial-temporal information aggregation in the frequency domain. Specifically, MFE mainly includes a spatial-frequency representation enhancement branch which captures the long-range dependency in the spatial dimension, and an energy frequency representation enhancement branch to obtain the inter-channel feature relationship. Moreover, a novel model training method named privilege training is proposed to encode the privilege information from high-resolution videos to facilitate model training. With these two methods, we introduce a new VSR model named MFPI, which outperforms state-of-the-art methods by a large margin while maintaining good efficiency on various datasets, including REDS4, Vimeo, Vid4, and UDM10.

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