检索结果-内蒙古大学图书馆

UNIFORMER: UNIFIED TRANSFORMER FOR EFFICIENT SPATIOTEMPORAL REPRESENTATION LEARNING 10

学校读者我要写书评

暂无评论

UNIFORMER: UNIFIED TRANSFORMER FOR EFFICIENT SPATIOTEMPORAL ...

10th International Conference on Learning Representations, ICLR 2022

作者： Li, Kunchang Wang, Yali Gao, Peng Song, Guanglu Liu, Yu Li, Hongsheng Qiao, Yu ShenZhen Key Lab of Computer Vision and Pattern Recognition SIAT-SenseTime Joint Lab Shenzhen Institute of Advanced Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China Shanghai AI Laboratory Shanghai China SenseTime Research The Chinese University of Hong Kong Hong Kong

It is a challenging task to learn rich and multi-scale spatiotemporal semantics from high-dimensional videos, due to large local redundancy and complex global dependency between video frames. The recent advances in this research have been mainly driven by 3D convolutional neural networks and vision transformers. Although 3D convolution can efficiently aggregate local context to suppress local redundancy from a small 3D neighborhood, it lacks the capability to capture global dependency because of the limited receptive field. Alternatively, vision transformers can effectively capture long-range dependency by self-attention mechanism, while having the limitation on reducing local redundancy with blind similarity comparison among all the tokens in each layer. Based on these observations, we propose a novel Unified transFormer (UniFormer) which seamlessly integrates merits of 3D convolution and spatiotemporal self-attention in a concise transformer format, and achieves a preferable balance between computation and accuracy. Different from traditional transformers, our relation aggregator can tackle both spatiotemporal redundancy and dependency, by learning local and global token affinity respectively in shallow and deep layers. We conduct extensive experiments on the popular video benchmarks, e.g., Kinetics-400, Kinetics-600, and Something-Something V1&V2. With only ImageNet-1K pretraining, our UniFormer achieves 82.9%/84.8% top-1 accuracy on Kinetics-400/Kinetics-600, while requiring 10× fewer GFLOPs than other state-of-the-art methods. For Something-Something V1 and V2, our UniFormer achieves new state-of-the-art performances of 60.9% and 71.2% top-1 accuracy respectively. Code is available at https://***/Sense-X/UniFormer. © 2022 ICLR 2022 - 10th International Conference on Learning Representationss. All rights reserved.

关键词： Redundancy

Fast Candidate Region Extraction for SAR Ship Target 37

学校读者我要写书评

暂无评论

Fast Candidate Region Extraction for SAR Ship Target

37th Youth Academic Annual Conference of Chinese Association of Automation, YAC 2022

作者： Zhang, Panpan Luo, Haibo Xu, Zheng He, Miao Shenyang Institute of Automation Shenyang110016 China Institutes for Robotics and Intelligent Manufacturing Shenyang110016 China University of Chinese Academy of Sciences Beijing100049 China Key Laboratory of Opto-Electronic Information Processing Shenyang110016 China The Key Lab of Image Understanding and Computer Vision Shenyang110016 China

ISBN: (纸本)9781665465366

At present, deep learning technology is widely used in ship target detection in synthetic aperture radar (SAR) images. However, high-resolution remote sensing SAR images cover a larger area and have larger image sizes. To be able to use the deep learning model for training and testing, the image needs to be cropped to the appropriate size. In a high-resolution SAR ship image, the ship target usually takes only a small part of the whole image. As a result, only part of the cropped image contains the target, and most of the rest are background regions. This will cause a lot of computational redundancy in the model inference stage. To solve this problem, a fast candidate region extraction algorithm is proposed in this paper for ship target extraction. The algorithm consists of three parts: firstly, extract the saliency map of the SAR ship image, secondly, perform median filtering on the segmented image, and thirdly, extract candidate regions. The superiority of the algorithm was demonstrated by experiments on the AIR-SARShip-1.0 dataset. © 2022 IEEE.

关键词： Synthetic aperture radar

PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation

学校读者我要写书评

暂无评论

PoseFormerV2: Exploring Frequency Domain for Efficient and R...

Conference on computer vision and Pattern Recognition (CVPR)

作者： Qitao Zhao Ce Zheng Mengyuan Liu Pichao Wang Chen Chen Shandong University Key Laboratory of Machine Perception Peking University Shenzhen Graduate School Amazon Prime Video Center for Research in Computer Vision University of Central Florida

Recently, transformer-based methods have gained significant success in sequential 2D-to-3D lifting human pose estimation. As a pioneering work, PoseFormer captures spatial relations of human joints in each video frame and human dynamics across frames with cascaded transformer layers and has achieved impressive performance. However, in real scenarios, the performance of PoseFormer and its follow-ups is limited by two factors: (a) The length of the input joint sequence; (b) The quality of 2D joint detection. Existing methods typically apply self-attention to all frames of the input sequence, causing a huge computational burden when the frame number is increased to obtain advanced estimation accuracy, and they are not robust to noise naturally brought by the limited capability of 2D joint detectors. In this paper, we propose PoseFormerV2, which exploits a compact representation of lengthy skeleton sequences in the frequency domain to efficiently scale up the receptive field and boost robustness to noisy 2D joint detection. With minimum modifications to PoseFormer, the proposed method effectively fuses features both in the time domain and frequency domain, enjoying a better speed-accuracy trade-off than its precursor. Extensive experiments on two benchmark datasets (i.e., Human3.6M and MPI-INF-3DHP) demonstrate that the proposed approach significantly outperforms the original PoseFormer and other transformer-based variants. Code is released at https://***/ QitaoZhao/PoseFormerV2.

关键词：

Which One is Better? Self-supervised Temporal Coherence Learning for Skeleton Based Action Recognition

学校读者我要写书评

暂无评论

Which One is Better? Self-supervised Temporal Coherence Lear...

IEEE International Joint Conference on Biometrics (IJCB)

作者： Bizhu Wu Mingyan Wu Haoqin Ji Linlin Shen Computer Vision Institute College of Computer Science and Software Engineering Guangdong Key Laboratory of Intelligent Information Processing Shenzhen University China

ISBN: (纸本)9781665463959

Recently, researchers have achieved significant results in the skeleton based action recognition task. To better model the skeleton sequences, existing methods learned the feature representations in the self-supervised setting by solving pretext tasks, such as predicting the order of a shuffled skeleton sequence or verifying whether a given skeleton sequence is shuffled or not. However, these pretext tasks are either too challenging or too easy for the encoder to obtain a proper skeleton representation for action recognition. Therefore, we propose a novel self-pretraining pretext task, Which One Is Better (WOIB), to identify which one is more temporally coherent, given two shuffled skeleton sequences. Experiments on the NTU RGB+D, NTU RGB+D 120, and Kinetics-Skeleton datasets with different network architectures show significant improvements in recognition accuracy, demonstrating that such a well-designed pretext task is general and able to drive the encoder to learn more discriminative representations.

关键词： Biometrics (access control) Coherence Predictive models Network architecture Drives Benchmark testing Skeleton

ODD: Omni Differential Drive for Simultaneous Reconfiguration and Omnidirectional Mobility of Wheeled Robots

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Ziqi Xie, Peijia Meng, Max Q.-H. Shenzhen Key Laboratory of Robotics Perception and Intelligence The Department of Electronic and Electrical Engineering Southern University of Science and Technology Shenzhen China The Department of Electronic Engineering The Chinese University of Hong Kong Hong Kong The Department of Electrical and Computer Engineering The University of Alberta Canada

Wheeled robots are highly efficient in human living environments. However, conventional wheeled designs, with their limited degrees of freedom and constraints in robot configuration, struggle to simultaneously achieve stability, passability, and agility due to varying footprint needs. This paper proposes a novel robot drive model inspired by human movements, termed as the Omni Differential Drive (ODD). The ODD model innovatively utilizes a lateral differential drive to adjust wheel spacing without adding additional actuators to the existing omnidirectional drive. This approach enables wheeled robots to achieve both simultaneous reconfiguration and omnidirectional mobility. To validate the feasibility of the ODD model, a functional prototype was developed, followed by comprehensive kinematic analyses. Control systems for self-balancing and motion control were designed and implemented. Experimental validations confirmed the feasibility of the ODD mechanism and the effectiveness of the control strategies. The results underline the potential of this innovative drive system to enhance the mobility and adaptability of robotic platforms. Copyright © 2024, The Authors. All rights reserved.

关键词： Kinematics

MM-3DScene: 3D Scene Understanding by Customizing Masked Modeling with Informative-Preserved Reconstruction and Self-Distilled Consistency

学校读者我要写书评

暂无评论

MM-3DScene: 3D Scene Understanding by Customizing Masked Mod...

Conference on computer vision and Pattern Recognition (CVPR)

作者： Mingye Xu Mutian Xu Tong He Wanli Ouyang Yali Wang Xiaoguang Han Yu Qiao The Guangdong Provincial Key Laboratory of Computer Vision and Virtual Reality Technology Chinese Academy of Sciences Shenzhen Institute of Advanced Technology Shenzhen China University of Chinese Academy of Sciences Shanghai Artificial Intelligence Laboratory SSE CUHKSZ FNii CUHKSZ

Masked Modeling (MM) has demonstrated widespread success in various vision challenges, by reconstructing masked visual patches. Yet, applying MM for large-scale 3D scenes remains an open problem due to the data sparsity and scene complexity. The conventional random masking paradigm used in 2D images often causes a high risk of ambiguity when recovering the masked region of 3D scenes. To this end, we propose a novel informative-preserved reconstruction, which explores local statistics to discover and preserve the representative structured points, effectively enhancing the pretext masking task for 3D scene understanding. Integrated with a progressive reconstruction manner, our method can concentrate on modeling regional geometry and enjoy less ambiguity for masked reconstruction. Besides, such scenes with progressive masking ratios can also serve to self-distill their intrinsic spatial consistency, requiring to learn the consistent representations from unmasked areas. By elegantly combining informative-preserved reconstruction on masked areas and consistency self-distillation from unmasked areas, a unified framework called MM-3DScene is yielded. We conduct comprehensive experiments on a host of downstream tasks. The consistent improvement (e.g., +6.1% mAP@0.5 on object detection and +2.2% mIoU on semantic segmentation) demonstrates the superiority of our approach.

关键词：

ODD: Omni Differential Drive for Simultaneous Reconfiguration and Omnidirectional Mobility of Wheeled Robots

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Ziqi Xie, Peijia Meng, Max Q.-H. Shenzhen Key Laboratory of Robotics Perception and Intelligence The Department of Electronic and Electrical Engineering Southern University of Science and Technology Shenzhen China Department of Electronic Engineering The Chinese University of Hong Kong Hong Kong Department of Electrical and Computer Engineering The University of Alberta Canada

Wheeled robots are highly efficient in human living environments. However, conventional wheeled designs, limited by degrees of freedom, struggle to meet varying footprint needs and achieve omnidirectional mobility. This paper proposes a novel robot drive model inspired by human movements, termed as the Omni Differential Drive (ODD). The ODD model innovatively utilizes a lateral differential drive to adjust wheel spacing without adding additional actuators to the existing omnidirectional drive. This approach enables wheeled robots to achieve both simultaneous reconfiguration and omnidirectional mobility. Additionally, a prototype was developed to validate the ODD, followed by kinematic analysis. Control systems for self-balancing and motion were designed and implemented. Experimental validations confirmed the feasibility of the ODD mechanism and the effectiveness of the control strategies. The results underline the potential of this innovative drive system to enhance the mobility and adaptability of robotic platforms. Copyright © 2024, The Authors. All rights reserved.

关键词： Wheels

COLLABORATIVE AUTO-ENCODING FOR BLIND IMAGE QUALITY ASSESSMENT

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Zhou, Zehong Zhou, Fei Qiu, Guoping College of Electronic and Information Engineering Shenzhen University Shenzhen China Peng Cheng Laboratory Shenzhen China Guangdong Key Laboratory of Intelligent Information Processing Shenzhen China Shenzhen Institute for Artificial Intelligence and Robotics for Society Shenzhen China School of Computer Science University of Nottingham Nottingham United Kingdom Guangdong-Hong Kong Joint Laboratory for Big Data Imaging and Communication Shenzhen China

Blind image quality assessment (BIQA) is a challenging problem with important real-world applications. Recent efforts attempting to exploit powerful representations by deep neural networks (DNN) are hindered by the lack of subjectively annotated data. This paper presents a novel BIQA method which overcomes this fundamental obstacle. Specifically, we design a pair of collaborative autoencoders (COAE) consisting of a content autoencoder (CAE) and a distortion autoencoder (DAE) that work together to extract content and distortion representations, which are shown to be highly descriptive of image quality. While the CAE follows a standard codec procedure, we introduce the CAE-encoded feature as an extra input to the DAE’s decoder for reconstructing distorted images, thus effectively forcing DAE’s encoder to extract distortion representations. The self-supervised learning framework allows the COAE including two feature extractors to be trained by almost unlimited amount of data, thus leaving limited samples with annotations to finetune a BIQA model. We will show that the proposed BIQA method achieves state-of-the-art performance and has superior generalization capability over other learning based models. The codes are available at: https://***/Macro-Zhou/NRIQA-VISOR/. © 2023, CC BY.

关键词： Supervised learning

Uncertainty-Estimation with Normalized Logits for Out-of-Distribution Detection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Huang, Mouxiao Qiao, Yu The Guangdong Provincial Key Laboratory of Computer Vision and Virtual Reality Technology Shenzhen Institute of Advanced Technology Chinese Academy of Sciences China University of Chinese Academy of Sciences China

Out-of-distribution (OOD) detection is critical for preventing deep learning models from making incorrect predictions to ensure the safety of artificial intelligence systems. Especially in safety-critical applications such as medical diagnosis and autonomous driving, the cost of incorrect decisions is usually unbearable. However, neural networks often suffer from the overconfidence issue, making high confidence for OOD data which are never seen during training process and may be irrelevant to training data, namely in-distribution (ID) data. Determining the reliability of the prediction is still a difficult and challenging task. In this work, we propose Uncertainty-Estimation with Normalized Logits (UE-NL), a robust learning method for OOD detection, which has three main benefits. (1) Neural networks with UE-NL treat every ID sample equally by predicting the uncertainty score of input data and the uncertainty is added into softmax function to adjust the learning strength of easy and hard samples during training phase, making the model learn robustly and accurately. (2) UE-NL enforces a constant vector norm on the logits to decouple the effect of the increasing output’s norm from optimization process, which causes the overconfidence issue to some extent. (3) UE-NL provides a new metric, the magnitude of uncertainty score, to detect OOD data. Experiments demonstrate that UE-NL achieves top performance on common OOD benchmarks and is more robust to noisy ID data that may be misjudged as OOD data by other methods. Copyright © 2023, The Authors. All rights reserved.

关键词： Forecasting