检索结果-内蒙古大学图书馆

HyRSM++: Hybrid relation guided temporal set matching for few-shot action recognition

学校读者我要写书评

暂无评论

Pattern Recognition 2024年 147卷

作者： Wang, Xiang Zhang, Shiwei Qing, Zhiwu Zuo, Zhengrong Gao, Changxin Jin, Rong Sang, Nong Key Laboratory of Ministry of Education for Image Processing and Intelligent Control School of Artificial Intelligence and Automation Huazhong University of Science and Technology China Alibaba Group China Meta AI United States

Few-shot action recognition is a challenging but practical problem aiming to learn a model that can be easily adapted to identify new action categories with only a few labeled samples. However, existing attempts 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 bidirectional Mean Hausdorff Metric to improve the resilience to misaligned instances. 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 consistently outperforms existing methods and achieves state-of-the-art performance under various few-shot settings. The source code is available at https://***/alibaba-mmai-research/HyRSMPlusPlus. © 2023 Elsevier Ltd

关键词： Benchmarking

Semi-Supervised Learning for Low-Light image Enhancement by Pseudo Low-Light image 16

学校读者我要写书评

暂无评论

Semi-Supervised Learning for Low-Light Image Enhancement by ...

16th International Congress on image and Signal processing, BioMedical Engineering and Informatics, CISP-BMEI 2023

作者： Xie, Shuo Ma, Yingjie Xu, Wei Qiu, Song Sun, Yifei East China Normal University School of Communication and Electronic Engineering China Ministry of Education Key Laboratory of System Control and Information Processing Shanghai200240 China

ISBN: (纸本)9798350330755

Building paired datasets in low-light enhancement entails significant cost and time, making such datasets precious commodities. Many researchers focus on how to enable models to learn more information from limited datasets. A prevalent strategy involves employing semi-supervised learning techniques to enhance model performance through additional unpaired images. However, one of the main challenges faced is the scarcity of a vast number of unpaired images from the same domain as the original low-light images. Consequently, we introduce a semi-supervised image enhancement method using pseudo-low-light images. Initially, we generate pseudo low-light images with less noise compared with the source domain image by the Signal-to-Noise Ratio prior and diffusion models. We then employ the Mean-Teacher network and the feature constraints of the pseudo-low-light images to realize low-light image enhancement. Comprehensive experimental results validate the efficacy of our approach on real-world datasets. © 2023 IEEE.

关键词： image enhancement

Late Better than Early: A Decision-Level Information Fusion Approach for Rgb-Thermal Crowd Counting with Illumination Awareness

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Cheng, Jian Feng, Chen Xiao, Yang Cao, Zhiguo Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

In this paper, we make the first research effort to address the RGB-Thermal (RGB-T) crowd counting problem with decision-level late fusion manner. Being different from the existing pixel-level or feature-level fusion methods, our proposition chooses to fuse the density maps yielded by RGB and thermal counterparts via spatially adaptive weighting with RGB illumination-aware attention. Our key intuition to conduct RGB-T density map fusion lies in 3 main folders. First, compared with the raw RGB-T images or convolutional feature maps, RGB-T density maps contain stronger counting-wise semantic meanings. Secondly, they are also of high spatial resolution for revealing fine local details. Thirdly, RGB-T density maps are complementary to each other for effective counting pattern characterization. To fuse them adaptively, a spatial weighting map is generated concerning the issues of RGB illumination awareness and local counting pattern characterization ability jointly. To the best of our knowledge, we are the first to leverage RGB-T crowd counting via concerning these 2 issues in a unified way. Meanwhile, cross-modality feature interaction is conducted between RGB and thermal modalities to facilitate spatial weighting map generation. The experiments on 2 well-established RGB-T crowd counting datasets, i.e., RGBT-CC and DroneRGBT) verify the superiority of our proposition. The source code and pretrained models will be released upon acceptance at https://***/hustaia/DLF-IA. © 2024, The Authors. All rights reserved.

关键词： Semantics

Cross-Task Inconsistency Based Active Learning (CTIAL) for Emotion Recognition

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Yifan Jiang, Xue Wu, Dongrui Key Laboratory of the Ministry of Education for Image Processing and Intelligent Control School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

Emotion recognition is a critical component of affective computing. Training accurate machine learning models for emotion recognition typically requires a large amount of labeled data. Due to the subtleness and complexity of emotions, multiple evaluators are usually needed for each affective sample to obtain its ground-truth label, which is expensive. To save the labeling cost, this paper proposes an inconsistency-based active learning approach for cross-task transfer between emotion classification and estimation. Affective norms are utilized as prior knowledge to connect the label spaces of categorical and dimensional emotions. Then, the prediction inconsistency on the two tasks for the unlabeled samples is used to guide sample selection in active learning for the target task. Experiments on within-corpus and cross-corpus transfers demonstrated that cross-task inconsistency could be a very valuable metric in active learning. To our knowledge, this is the first work that utilizes prior knowledge on affective norms and data in a different task to facilitate active learning for a new task, even the two tasks are from different datasets. © 2024, CC BY.

关键词： Active learning

Spatiotemporal distance embedded hybrid ant colony algorithm for a kind of vehicle routing problem with constraints

学校读者我要写书评

暂无评论

Frontiers of Information Technology & Electronic Engineering 2023年第7期24卷 1062-1079页

作者： Zhenhui FENG Renbin XIAO School of Artificial Intelligence and Automation Huazhong University of Science and TechnologyWuhan 430074China Institute of Artificial Intelligence Huazhong University of Science and TechnologyWuhan 430074China Key Laboratory of Image Information Processing and Intelligent Control Ministry of EducationWuhan 430074China

We investigate a kind of vehicle routing problem with constraints(VRPC)in the car-sharing mobility environment,where the problem is based on user orders,and each order has a reservation time limit and two location point transitions,origin and *** is a typical extended vehicle routing problem(VRP)with both time and space *** consider the VRPC problem characteristics and establish a vehicle scheduling model to minimize operating costs and maximize user(or passenger)*** solve the scheduling model more accurately,a spatiotemporal distance representation function is defined based on the temporal and spatial properties of the customer,and a spatiotemporal distance embedded hybrid ant colony algorithm(HACA-ST)is *** algorithm can be divided into two ***,through spatiotemporal clustering,the spatiotemporal distance between users is the main measure used to classify customers in categories,which helps provide heuristic information for problem ***,an improved ant colony algorithm(ACO)is proposed to optimize the solution by combining a labor division strategy and the spatiotemporal distance function to obtain the final scheduling *** analysis is carried out based on existing data sets and simulated urban *** with other heuristic algorithms,HACA-ST reduces the length of the shortest route by 2%–14%in benchmark *** VRPC testing instances,concerning the combined cost,HACA-ST has competitive cost compared to existing VRP-related ***,we provide two actual urban scenarios to further verify the effectiveness of the proposed algorithm.

关键词： Vehicle routing problem with constraints(VRPC) Spatiotemporal distance function Labor division strategy Ant colony algorithm(ACO)

Pseudo-Labeling by Multi-Policy Viewfinder Network for image Cropping

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Pan, Zhiyu Wang, Kewei Wu, Yizheng Xiao, Liwen Cui, Jiahao Wang, Zhicheng Cao, Zhiguo Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

Automatic image cropping models predict reframing boxes to enhance image aesthetics. Yet, the scarcity of labeled data hinders the progress of this task. To overcome this limitation, we explore the possibility of utilizing both labeled and unlabeled data together to expand the scale of training data for image cropping models. This idea can be implemented in a pseudo-labeling way: producing pseudo labels for unlabeled data by a teacher model and training a student model with these pseudo labels. However, the student may learn from teacher's mistakes. To address this issue, we propose the multi-policy viewfinder network (MPV-Net) that offers diverse refining policies to rectify the mistakes in original pseudo labels from the teacher. The most reliable policy is selected to generate trusted pseudo labels. The reliability of policies is evaluated via the robustness against box jittering. The efficacy of our method can be evaluated by the improvement compared to the supervised baseline which only uses labeled data. Notably, our MPV-Net outperforms off-the-shelf pseudo-labeling methods, yielding the most substantial improvement over the supervised baseline. Furthermore, our approach achieves state-of-the-art results on both the FCDB and FLMS datasets, signifying the superiority of our approach. Copyright © 2024, The Authors. All rights reserved.

关键词： Distillation

FADE: A Task-Agnostic Upsampling Operator for Encoder-Decoder Architectures

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lu, Hao Liu, Wenze Fu, Hongtao Cao, Zhiguo The Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

The goal of this work is to develop a task-agnostic feature upsampling operator for dense prediction where the operator is required to facilitate not only region-sensitive tasks like semantic segmentation but also detail-sensitive tasks such as image matting. Prior upsampling operators often can work well in either type of the tasks, but not both. We argue that task-agnostic upsampling should dynamically trade off between semantic preservation and detail delineation, instead of having a bias between the two properties. In this paper, we present FADE, a novel, plug-and-play, lightweight, and task-agnostic upsampling operator by fusing the assets of decoder and encoder features at three levels: i) considering both the encoder and decoder feature in upsampling kernel generation;ii) controlling the per-point contribution of the encoder/decoder feature in upsampling kernels with an efficient semi-shift convolutional operator;and iii) enabling the selective pass of encoder features with a decoder-dependent gating mechanism for compensating details. To improve the practicality of FADE, we additionally study parameter- and memory-efficient implementations of semi-shift convolution. We analyze the upsampling behavior of FADE on toy data and show through large-scale experiments that FADE is task-agnostic with consistent performance improvement on a number of dense prediction tasks with little extra cost. For the first time, we demonstrate robust feature upsampling on both region- and detail-sensitive tasks successfully. Code is made available at: https://***/poppinace/fade Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

A3E: Aligned and Augmented Adversarial Ensemble for Accurate, Robust and Privacy-Preserving EEG Decoding

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Chen, Xiaoqing Jia, Tianwang Wu, Dongrui Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

An electroencephalogram (EEG) based brain-computer interface (BCI) enables direct communication between the brain and external devices. However, EEG-based BCIs face at least three major challenges in real-world applications: data scarcity and individual differences, adversarial vulnerability, and data privacy. While previous studies have addressed one or two of these issues, simultaneous accommodation of all three challenges remains challenging and unexplored. This paper fills this gap, by proposing an Aligned and Augmented Adversarial Ensemble (A3E) algorithm and integrating it into three privacy protection scenarios (centralized source-free transfer, federated source-free transfer, and source data perturbation), achieving simultaneously accurate decoding, adversarial robustness, and privacy protection of EEG-based BCIs. Experiments on three public EEG datasets demonstrated that our proposed approach outperformed over 10 classic and state-of-the-art approaches in both accuracy and robustness in all three privacy-preserving scenarios, even outperforming state-of-the-art transfer learning approaches that do not consider privacy protection at all. This is the first time that three major challenges in EEG-based BCIs can be addressed simultaneously, significantly improving the practicalness of EEG decoding in real-world BCIs. © 2024, CC BY.

关键词： Transfer learning

Knowledge-Data Fusion Based Source-Free Semi-Supervised Domain Adaptation for Seizure Subtype Classification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Peng, Ruimin An, Jiayu Wu, Dongrui Key Laboratory of the Ministry of Education for Image Processing and Intelligent Control School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China

Electroencephalogram (EEG)-based seizure subtype classification enhances clinical diagnosis efficiency. Source-free semi-supervised domain adaptation (SF-SSDA), which transfers a pre-trained model to a new dataset with no source data and limited labeled target data, can be used for privacy-preserving seizure subtype classification. This paper considers two challenges in SF-SSDA for EEG-based seizure subtype classification: 1) How to effectively fuse both raw EEG data and expert knowledge in classifier design? 2) How to align the source and target domain distributions for SF-SSDA? We propose a Knowledge-Data Fusion based SF-SSDA approach, KDF-MutualSHOT, for EEG-based seizure subtype classification. In source model training, KDF uses Jensen-Shannon Divergence to facilitate mutual learning between a feature-driven Decision Tree-based model and a data-driven Transformer-based model. To adapt KDF to a new target dataset, an SF-SSDA algorithm, MutualSHOT, is developed, which features a consistency-based pseudo-label selection strategy. Experiments on the public TUSZ and CHSZ datasets demonstrated that KDF-MutualSHOT outperformed other supervised and source-free domain adaptation approaches in cross-subject seizure subtype classification. © 2024, CC BY.

关键词： Self-supervised learning