What representation do deep neural networks learn? How similar are images to each other for neural networks? Despite the overwhelming success of deep learning methods key questions about their internal workings still remain largely unanswered, due to their internal high dimensionality and complexity. To address this, one approach is to measure the similarity of activation responses to various inputs. Representational Similarity Matrices (RSMs) distill this similarity into scalar values for each input pair. These matrices encapsulate the entire similarity structure of a system, indicating which input leads to similar responses. While the similarity between images is ambiguous, we argue that the spatial location of semantic objects does neither influence human perception nor deep learning classifiers. Thus this should be reflected in the definition of similarity between image responses for computer vision systems. Revisiting the established similarity calculations for RSMs we expose their sensitivity to spatial alignment. In this paper, we propose to solve this through semantic RSMs, which are invariant to spatial permutation. We measure semantic similarity between input responses by formulating it as a set-matching problem. Further, we quantify the superiority of semantic RSMs over spatio-semantic RSMs through image retrieval and by comparing the similarity between representations to the similarity between predicted class probabilities.
Learning general-purpose models from diverse datasets has achieved great success in machine learning. In robotics, however, existing methods in multi-task learning are typically constrained to a single robot and workspace, while recent work such as RT-X requires a non-trivial action normalization procedure to manually bridge the gap between different action spaces in diverse environments. In this paper, we propose the visual kinematics chain as a precise and universal representation of quasi-static actions for robot learning over diverse environments, which requires no manual adjustment since the visual kinematic chains can be automatically obtained from the robot's model and camera parameters. We propose the Visual Kinematics Transformer (VKT), a convolution-free architecture that supports an arbitrary number of camera viewpoints, and that is trained with a single objective of forecasting kinematic structures through optimal point-set matching. We demonstrate the superior performance of VKT over BC transformers as a general agent on Calvin, RLBench, Open-X, and real robot manipulation tasks. Video demonstrations can be found at https://mlzxy.github.io/visual-kinetic-chain.
We introduce Sketch-based Video Object Localization (SVOL), a new task aimed at localizing spatio-temporal object boxes in video queried by the input sketch. We first outline the challenges in the SVOL task and build the Sketch-Video Attention Network (SVANet) with the following design principles: (i) to consider temporal information of video and bridge the domain gap between sketch and video; (ii) to accurately identify and localize multiple objects simultaneously; (iii) to handle various styles of sketches; (iv) to be classification-free. In particular, SVANet is equipped with a Cross-modal Transformer that models the interaction between learnable object tokens, query sketch, and video through attention operations, and learns upon a per-frame set matching strategy that enables frame-wise prediction while utilizing global video context. We evaluate SVANet on a newly curated SVOL dataset. By design, SVANet successfully learns the mapping between the query sketches and video objects, achieving state-of-the-art results on the SVOL benchmark. We further confirm the effectiveness of SVANet via extensive ablation studies and visualizations. Lastly, we demonstrate its transfer capability on unseen datasets and novel categories, suggesting its high scalability in real-world applications.
Room layout estimation is a long-existing robotic vision task that benefits both environment sensing and motion planning. However, layout estimation using point clouds (PCs) still suffers from data scarcity due to annotation difficulty. As such, we address the semi-supervised setting of this task based upon the idea of model exponential moving averaging. But adapting this scheme to the state-of-the-art (SOTA) solution for PC-based layout estimation is not straightforward. To this end, we define a quad set matching strategy and several consistency losses based upon metrics tailored for layout quads. Besides, we propose a new online pseudo-label harvesting algorithm that decomposes the distribution of a hybrid distance measure between quads and PC into two components. This technique does not need manual threshold selection and intuitively encourages quads to align with reliable layout points. Surprisingly, this framework also works for the fully-supervised setting, achieving a new SOTA on the ScanNet benchmark. Last but not least, we also push the semi-supervised setting to the realistic omni-supervised setting, demonstrating significantly promoted performance on a newly annotated ARKitScenes testing set. Our codes, data and models are released in this repository.
Recent attempts mainly focus on learning deep representations for each video individually under the episodic meta-learning regime and then performing temporal alignment to match query and support videos. However, they still suffer from two drawbacks: (i) learning individual features without considering the entire task may result in limited representation capability, and (ii) existing alignment strategies are sensitive to noises and misaligned instances. To handle the two limitations, we propose a novel Hybrid Relation guided temporal Set Matching (HyRSM++) approach for few-shot action recognition. The core idea of HyRSM++ is to integrate all videos within the task to learn discriminative representations and involve a robust matching technique. To be specific, HyRSM++ consists of two key components, a hybrid relation module and a temporal set matching metric. Given the basic representations from the feature extractor, the hybrid relation module is introduced to fully exploit associated relations within and cross videos in an episodic task and thus can learn task-specific embeddings. Subsequently, in the temporal set matching metric, we carry out the distance measure between query and support videos from a set matching perspective and design a Bi-MHM to improve the resilience to misaligned instances. In addition, we explicitly exploit the temporal coherence in videos to regularize the matching process. Furthermore, we extend the proposed HyRSM++ to deal with the more challenging semi-supervised few-shot action recognition and unsupervised few-shot action recognition tasks. Experimental results on multiple benchmarks demonstrate that our method achieves state-of-the-art performance under various few-shot settings. The source code is available at https://github.com/alibaba-mmai-research/HyRSMPlusPlus.
Existing object detection methods are bounded in a fixed-set vocabulary by costly labeled data. When dealing with novel categories, the model has to be retrained with more bounding box annotations. Natural language supervision is an attractive alternative for its annotation-free attributes and broader object concepts. However, learning open-vocabulary object detection from language is challenging since image-text pairs do not contain fine-grained object-language alignments. Previous solutions rely on either expensive grounding annotations or distilling classification-oriented vision models. In this paper, we propose a novel open-vocabulary object detection framework directly learning from image-text pair data. We formulate object-language alignment as a set matching problem between a set of image region features and a set of word embeddings. It enables us to train an open-vocabulary object detector on image-text pairs in a much simple and effective way. Extensive experiments on two benchmark datasets, COCO and LVIS, demonstrate our superior performance over the competing approaches on novel categories, e.g. achieving 32.0% mAP on COCO and 21.7% mask mAP on LVIS. Code is available at: https://github.com/clin1223/VLDet.
One-to-one set matching is a key design for DETR to establish its end-to-end capability, so that object detection does not require a hand-crafted NMS (non-maximum suppression) to remove duplicate detections. This end-to-end signature is important for the versatility of DETR, and it has been generalized to broader vision tasks. However, we note that there are few queries assigned as positive samples and the one-to-one set matching significantly reduces the training efficacy of positive samples. We propose a simple yet effective method based on a hybrid matching scheme that combines the original one-to-one matching branch with an auxiliary one-to-many matching branch during training. Our hybrid strategy has been shown to significantly improve accuracy. In inference, only the original one-to-one match branch is used, thus maintaining the end-to-end merit and the same inference efficiency of DETR. The method is named H-DETR, and it shows that a wide range of representative DETR methods can be consistently improved across a wide range of visual tasks, including DeformableDETR, PETRv2, PETR, and TransTrack, among others. The code is available at: https://github.com/HDETR
Current few-shot action recognition methods reach impressive performance by learning discriminative features for each video via episodic training and designing various temporal alignment strategies. Nevertheless, they are limited in that (a) learning individual features without considering the entire task may lose the most relevant information in the current episode, and (b) these alignment strategies may fail in misaligned instances. To overcome the two limitations, we propose a novel Hybrid Relation guided Set Matching (HyRSM) approach that incorporates two key components: hybrid relation module and set matching metric. The purpose of the hybrid relation module is to learn task-specific embeddings by fully exploiting associated relations within and cross videos in an episode. Built upon the task-specific features, we reformulate distance measure between query and support videos as a set matching problem and further design a bidirectional Mean Hausdorff Metric to improve the resilience to misaligned instances. By this means, the proposed HyRSM can be highly informative and flexible to predict query categories under the few-shot settings. We evaluate HyRSM on six challenging benchmarks, and the experimental results show its superiority over the state-of-the-art methods by a convincing margin. Project page: https://hyrsm-cvpr2022.github.io/.
Usual relations between entities could be captured using graphs; but those of a higher-order -- more so between two different types of entities (which we term "left" and "right") -- calls for a "bipartite hypergraph". For example, given a left set of symptoms and right set of diseases, the relation between a set subset of symptoms (that a patient experiences at a given point of time) and a subset of diseases (that he/she might be diagnosed with) could be well-represented using a bipartite hyperedge. The state-of-the-art in embedding nodes of a hypergraph is based on learning the self-attention structure between node-pairs from a hyperedge. In the present work, given a bipartite hypergraph, we aim at capturing relations between node pairs from the cross-product between the left and right hyperedges, and term it a "cross-attention" (CAT) based model. More precisely, we pose "bipartite hyperedge link prediction" as a set-matching (SETMAT) problem and propose a novel neural network architecture called CATSETMAT for the same. We perform extensive experiments on multiple bipartite hypergraph datasets to show the superior performance of CATSETMAT, which we compare with multiple techniques from the state-of-the-art. Our results also elucidate information flow in self- and cross-attention scenarios.
This paper addresses the problem of set-to-set matching, which involves matching two different sets of items based on some criteria, especially in the case of high-dimensional items like images. Although neural networks have been applied to solve this problem, most machine learning-based approaches assume that the training and test data follow the same distribution, which is not always true in real-world scenarios. To address this limitation, we introduce SHIFT15M, a dataset that can be used to evaluate set-to-set matching models when the distribution of data changes between training and testing. We conduct benchmark experiments that demonstrate the performance drop of naive methods due to distribution shift. Additionally, we provide software to handle the SHIFT15M dataset in a simple manner, with the URL for the software to be made available after publication of this manuscript. We believe proposed SHIFT15M dataset provide a valuable resource for evaluating set-to-set matching models under the distribution shift.
A common heuristic in semi-supervised deep learning (SSDL) is to select unlabelled data based on a notion of semantic similarity to the labelled data. For example, labelled images of numbers should be paired with unlabelled images of numbers instead of, say, unlabelled images of cars. We refer to this practice as semantic data set matching. In this work, we demonstrate the limits of semantic data set matching. We show that it can sometimes even degrade the performance for a state of the art SSDL algorithm. We present and make available a comprehensive simulation sandbox, called non-IID-SSDL, for stress testing an SSDL algorithm under different degrees of distribution mismatch between the labelled and unlabelled data sets. In addition, we demonstrate that simple density based dissimilarity measures in the feature space of a generic classifier offer a promising and more reliable quantitative matching criterion to select unlabelled data before SSDL training.
Finding an optimal correspondence between point sets is a common task in computer vision. Existing techniques assume relatively simple relationships among points and do not guarantee an optimal match. We introduce an algorithm capable of exactly solving point set matching by modeling the task as hypergraph matching. The algorithm extends the classical branch and bound paradigm to select and aggregate vertices under a proposed decomposition of the multilinear objective function. The methodology is motivated by Caenorhabditis elegans, a model organism used frequently in developmental biology and neurobiology. The embryonic C. elegans contains seam cells that can act as fiducial markers allowing the identification of other nuclei during embryo development. The proposed algorithm identifies seam cells more accurately than established point-set matching methods, while providing a framework to approach other similarly complex point set matching tasks.
Matching two different sets of items, called heterogeneous set-to-set matching problem, has recently received attention as a promising problem. The difficulties are to extract features to match a correct pair of different sets and also preserve two types of exchangeability required for set-to-set matching: the pair of sets, as well as the items in each set, should be exchangeable. In this study, we propose a novel deep learning architecture to address the abovementioned difficulties and also an efficient training framework for set-to-set matching. We evaluate the methods through experiments based on two industrial applications: fashion set recommendation and group re-identification. In these experiments, we show that the proposed method provides significant improvements and results compared with the state-of-the-art methods, thereby validating our architecture for the heterogeneous set matching problem.
Graph similarity computation is one of the core operations in many graph-based applications, such as graph similarity search, graph database analysis, graph clustering, etc. Since computing the exact distance/similarity between two graphs is typically NP-hard, a series of approximate methods have been proposed with a trade-off between accuracy and speed. Recently, several data-driven approaches based on neural networks have been proposed, most of which model the graph-graph similarity as the inner product of their graph-level representations, with different techniques proposed for generating one embedding per graph. However, using one fixed-dimensional embedding per graph may fail to fully capture graphs in varying sizes and link structures, a limitation that is especially problematic for the task of graph similarity computation, where the goal is to find the fine-grained difference between two graphs. In this paper, we address the problem of graph similarity computation from another perspective, by directly matching two sets of node embeddings without the need to use fixed-dimensional vectors to represent whole graphs for their similarity computation. The model, GraphSim, achieves the state-of-the-art performance on four real-world graph datasets under six out of eight settings (here we count a specific dataset and metric combination as one setting), compared to existing popular methods for approximate Graph Edit Distance (GED) and Maximum Common Subgraph (MCS) computation.