School of Computer Science and Engineering, University of Electronic Science and Technology of China
Abstract:The detection of moving infrared dim-small targets has been a challenging and prevalent research topic. The current state-of-the-art methods are mainly based on ConvLSTM to aggregate information from adjacent frames to facilitate the detection of the current frame. However, these methods implicitly utilize motion information only in the training stage and fail to explicitly explore motion compensation, resulting in poor performance in the case of a video sequence including large motion. In this paper, we propose a Deformable Feature Alignment and Refinement (DFAR) method based on deformable convolution to explicitly use motion context in both the training and inference stages. Specifically, a Temporal Deformable Alignment (TDA) module based on the designed Dilated Convolution Attention Fusion (DCAF) block is developed to explicitly align the adjacent frames with the current frame at the feature level. Then, the feature refinement module adaptively fuses the aligned features and further aggregates useful spatio-temporal information by means of the proposed Attention-guided Deformable Fusion (AGDF) block. In addition, to improve the alignment of adjacent frames with the current frame, we extend the traditional loss function by introducing a new motion compensation loss. Extensive experimental results demonstrate that the proposed DFAR method achieves the state-of-the-art performance on two benchmark datasets including DAUB and IRDST.
Abstract:Few-shot Semantic Segmentation (FSS) aims to adapt a pretrained model to new classes with as few as a single labelled training sample per class. Despite the prototype based approaches have achieved substantial success, existing models are limited to the imaging scenarios with considerably distinct objects and not highly complex background, e.g., natural images. This makes such models suboptimal for medical imaging with both conditions invalid. To address this problem, we propose a novel Detail Self-refined Prototype Network (DSPNet) to constructing high-fidelity prototypes representing the object foreground and the background more comprehensively. Specifically, to construct global semantics while maintaining the captured detail semantics, we learn the foreground prototypes by modelling the multi-modal structures with clustering and then fusing each in a channel-wise manner. Considering that the background often has no apparent semantic relation in the spatial dimensions, we integrate channel-specific structural information under sparse channel-aware regulation. Extensive experiments on three challenging medical image benchmarks show the superiority of DSPNet over previous state-of-the-art methods.
Abstract:Moving infrared small target detection presents significant challenges due to tiny target sizes and low contrast against backgrounds. Currently-existing methods primarily focus on extracting target features only from the spatial-temporal domain. For further enhancing feature representation, more information domains such as frequency are believed to be potentially valuable. To extend target feature learning, we propose a new Triple-domain Strategy (Tridos) with the frequency-aware memory enhancement on the spatial-temporal domain. In our scheme, it effectively detaches and enhances frequency features by a local-global frequency-aware module with Fourier transform. Inspired by the human visual system, our memory enhancement aims to capture the target spatial relations between video frames. Furthermore, it encodes temporal dynamics motion features via differential learning and residual enhancing. Additionally, we further design a residual compensation unit to reconcile possible cross-domain feature mismatches. To our best knowledge, our Tridos is the first work to explore target feature learning comprehensively in spatial-temporal-frequency domains. The extensive experiments on three datasets (DAUB, ITSDT-15K, and IRDST) validate that our triple-domain learning scheme could be obviously superior to state-of-the-art ones. Source codes are available at https://github.com/UESTC-nnLab/Tridos.
Abstract:Source-free Domain Adaptation (SFDA) aims to adapt a pre-trained source model to an unlabeled target domain with no access to the source data. Inspired by the success of pre-trained large vision-language (ViL) models in many other applications, the latest SFDA methods have also validated the benefit of ViL models by leveraging their predictions as pseudo supervision. However, we observe that ViL's predictions could be noisy and inaccurate at an unknown rate, potentially introducing additional negative effects during adaption. To address this thus-far ignored challenge, in this paper, we introduce a novel Proxy Denoising (ProDe) approach. Specifically, we leverage the ViL model as a proxy to facilitate the adaptation process towards the latent domain-invariant space. Critically, we design a proxy denoising mechanism for correcting ViL's predictions. This is grounded on a novel proxy confidence theory by modeling elegantly the domain adaption effect of the proxy's divergence against the domain-invariant space. To capitalize the corrected proxy, we further derive a mutual knowledge distilling regularization. Extensive experiments show that our ProDe significantly outperforms the current state-of-the-art alternatives under both conventional closed-set setting and the more challenging open-set, partial-set and generalized SFDA settings. The code will release soon.
Abstract:In the pursuit of transferring a source model to a target domain without access to the source training data, Source-Free Domain Adaptation (SFDA) has been extensively explored across various scenarios, including closed-set, open-set, partial-set, and generalized settings. Existing methods, focusing on specific scenarios, not only address only a subset of challenges but also necessitate prior knowledge of the target domain, significantly limiting their practical utility and deployability. In light of these considerations, we introduce a more practical yet challenging problem, termed unified SFDA, which comprehensively incorporates all specific scenarios in a unified manner. To tackle this unified SFDA problem, we propose a novel approach called Latent Causal Factors Discovery (LCFD). In contrast to previous alternatives that emphasize learning the statistical description of reality, we formulate LCFD from a causality perspective. The objective is to uncover the causal relationships between latent variables and model decisions, enhancing the reliability and robustness of the learned model against domain shifts. To integrate extensive world knowledge, we leverage a pre-trained vision-language model such as CLIP. This aids in the formation and discovery of latent causal factors in the absence of supervision in the variation of distribution and semantics, coupled with a newly designed information bottleneck with theoretical guarantees. Extensive experiments demonstrate that LCFD can achieve new state-of-the-art results in distinct SFDA settings, as well as source-free out-of-distribution generalization.Our code and data are available at https://github.com/tntek/source-free-domain-adaptation.
Abstract:Source-Free Domain Adaptation (SFDA) aims to adapt a source model for a target domain, with only access to unlabeled target training data and the source model pre-trained on a supervised source domain. Relying on pseudo labeling and/or auxiliary supervision, conventional methods are inevitably error-prone. To mitigate this limitation, in this work we for the first time explore the potentials of off-the-shelf vision-language (ViL) multimodal models (e.g.,CLIP) with rich whilst heterogeneous knowledge. We find that directly applying the ViL model to the target domain in a zero-shot fashion is unsatisfactory, as it is not specialized for this particular task but largely generic. To make it task specific, we propose a novel Distilling multimodal Foundation model(DIFO)approach. Specifically, DIFO alternates between two steps during adaptation: (i) Customizing the ViL model by maximizing the mutual information with the target model in a prompt learning manner, (ii) Distilling the knowledge of this customized ViL model to the target model. For more fine-grained and reliable distillation, we further introduce two effective regularization terms, namely most-likely category encouragement and predictive consistency. Extensive experiments show that DIFO significantly outperforms the state-of-the-art alternatives. Our source code will be released.
Abstract:Learned video compression (LVC) has witnessed remarkable advancements in recent years. Similar as the traditional video coding, LVC inherits motion estimation/compensation, residual coding and other modules, all of which are implemented with neural networks (NNs). However, within the framework of NNs and its training mechanism using gradient backpropagation, most existing works often struggle to consistently generate stable motion information, which is in the form of geometric features, from the input color features. Moreover, the modules such as the inter-prediction and residual coding are independent from each other, making it inefficient to fully reduce the spatial-temporal redundancy. To address the above problems, in this paper, we propose a novel Spatial-Temporal Transformer based Video Compression (STT-VC) framework. It contains a Relaxed Deformable Transformer (RDT) with Uformer based offsets estimation for motion estimation and compensation, a Multi-Granularity Prediction (MGP) module based on multi-reference frames for prediction refinement, and a Spatial Feature Distribution prior based Transformer (SFD-T) for efficient temporal-spatial joint residual compression. Specifically, RDT is developed to stably estimate the motion information between frames by thoroughly investigating the relationship between the similarity based geometric motion feature extraction and self-attention. MGP is designed to fuse the multi-reference frame information by effectively exploring the coarse-grained prediction feature generated with the coded motion information. SFD-T is to compress the residual information by jointly exploring the spatial feature distributions in both residual and temporal prediction to further reduce the spatial-temporal redundancy. Experimental results demonstrate that our method achieves the best result with 13.5% BD-Rate saving over VTM.
Abstract:To solve the problem of poor performance of deep neural network models due to insufficient data, a simple yet effective interpolation-based data augmentation method is proposed: MSMix (Manifold Swap Mixup). This method feeds two different samples to the same deep neural network model, and then randomly select a specific layer and partially replace hidden features at that layer of one of the samples by the counterpart of the other. The mixed hidden features are fed to the model and go through the rest of the network. Two different selection strategies are also proposed to obtain richer hidden representation. Experiments are conducted on three Chinese intention recognition datasets, and the results show that the MSMix method achieves better results than other methods in both full-sample and small-sample configurations.
Abstract:Balancing efficiency and accuracy is a long-standing problem for deploying deep learning models. The trade-off is even more important for real-time safety-critical systems like autonomous vehicles. In this paper, we propose an effective approach for accelerating transformer-based 3D object detectors by dynamically halting tokens at different layers depending on their contribution to the detection task. Although halting a token is a non-differentiable operation, our method allows for differentiable end-to-end learning by leveraging an equivalent differentiable forward-pass. Furthermore, our framework allows halted tokens to be reused to inform the model's predictions through a straightforward token recycling mechanism. Our method significantly improves the Pareto frontier of efficiency versus accuracy when compared with the existing approaches. By halting tokens and increasing model capacity, we are able to improve the baseline model's performance without increasing the model's latency on the Waymo Open Dataset.
Abstract:In this paper, we consider the problem of bit allocation in neural video compression (NVC). Due to the frame reference structure, current NVC methods using the same R-D (Rate-Distortion) trade-off parameter $\lambda$ for all frames are suboptimal, which brings the need for bit allocation. Unlike previous methods based on heuristic and empirical R-D models, we propose to solve this problem by gradient-based optimization. Specifically, we first propose a continuous bit implementation method based on Semi-Amortized Variational Inference (SAVI). Then, we propose a pixel-level implicit bit allocation method using iterative optimization by changing the SAVI target. Moreover, we derive the precise R-D model based on the differentiable trait of NVC. And we show the optimality of our method by proofing its equivalence to the bit allocation with precise R-D model. Experimental results show that our approach significantly improves NVC methods and outperforms existing bit allocation methods. Our approach is plug-and-play for all differentiable NVC methods, and it can be directly adopted on existing pre-trained models.