1 code implementation • 19 Jul 2022 • Li Zhang, Sixiao Zheng, Jiachen Lu, Xinxuan Zhao, Xiatian Zhu, Yanwei Fu, Tao Xiang, Jianfeng Feng
Extensive experiments show that our method achieves appealing performance on a variety of visual recognition tasks (e. g., image classification, object detection and instance segmentation and semantic segmentation).
1 code implementation • 17 Jul 2022 • Xiao Han, Licheng Yu, Xiatian Zhu, Li Zhang, Yi-Zhe Song, Tao Xiang
We thus propose a Multi-View Contrastive Learning task for pulling closer the visual representation of one image to the compositional multimodal representation of another image+text.
1 code implementation • 17 Jul 2022 • Sauradip Nag, Xiatian Zhu, Yi-Zhe Song, Tao Xiang
Such a novel design effectively eliminates the dependence between localization and classification by breaking the route for error propagation in-between.
Ranked #1 on
Zero-Shot Action Detection
on ActivityNet-1.3
2 code implementations • 14 Jul 2022 • Sauradip Nag, Xiatian Zhu, Yi-Zhe Song, Tao Xiang
Existing temporal action detection (TAD) methods rely on generating an overwhelmingly large number of proposals per video.
Ranked #7 on
Temporal Action Localization
on THUMOS’14
1 code implementation • 14 Jul 2022 • Sauradip Nag, Xiatian Zhu, Yi-Zhe Song, Tao Xiang
Such a novel design effectively eliminates the dependence between localization and classification by cutting off the route for error propagation in-between.
Ranked #1 on
Semi-Supervised Action Detection
on THUMOS' 14
1 code implementation • 5 Jul 2022 • Jiachen Lu, Li Zhang, Junge Zhang, Xiatian Zhu, Hang Xu, Jianfeng Feng
Crucially, with a linear complexity, much longer token sequences are permitted in SOFT, resulting in superior trade-off between accuracy and complexity.
1 code implementation • 5 Jul 2022 • Hengyuan Ma, Li Zhang, Xiatian Zhu, Jianfeng Feng
However, a fundamental limitation is that their inference is very slow due to a need for many (e. g., 2000) iterations of sequential computations.
1 code implementation • 30 Jun 2022 • Yanqin Jiang, Li Zhang, Zhenwei Miao, Xiatian Zhu, Jin Gao, Weiming Hu, Yu-Gang Jiang
3D object detection in autonomous driving aims to reason "what" and "where" the objects of interest present in a 3D world.
no code implementations • 27 Jun 2022 • Junting Pan, Ziyi Lin, Xiatian Zhu, Jing Shao, Hongsheng Li
However, existing attempts typically focus on downstream tasks from the same modality (e. g., image understanding) of the pre-trained model.
Ranked #6 on
Action Classification
on Kinetics-400
(using extra training data)
no code implementations • 13 Jun 2022 • Peng Xu, Xiatian Zhu, David A. Clifton
Transformer is a promising neural network learner, and has achieved great success in various machine learning tasks.
1 code implementation • 8 Jun 2022 • Jiachen Lu, Zheyuan Zhou, Xiatian Zhu, Hang Xu, Li Zhang
A self-driving perception model aims to extract 3D semantic representations from multiple cameras collectively into the bird's-eye-view (BEV) coordinate frame of the ego car in order to ground downstream planner.
no code implementations • 8 Jun 2022 • Hengyuan Ma, Li Zhang, Xiatian Zhu, Jingfeng Zhang, Jianfeng Feng
To ensure stability of convergence in sampling and generation quality, however, this sequential sampling process has to take a small step size and many sampling iterations (e. g., 2000).
1 code implementation • 13 May 2022 • Jing Yang, Xiatian Zhu, Adrian Bulat, Brais Martinez, Georgios Tzimiropoulos
The key idea is that we leverage the teacher's classifier as a semantic critic for evaluating the representations of both teacher and student and distilling the semantic knowledge with high-order structured information over all feature dimensions.
2 code implementations • 6 May 2022 • Junting Pan, Adrian Bulat, Fuwen Tan, Xiatian Zhu, Lukasz Dudziak, Hongsheng Li, Georgios Tzimiropoulos, Brais Martinez
In this work, pushing further along this under-studied direction we introduce EdgeViTs, a new family of light-weight ViTs that, for the first time, enable attention-based vision models to compete with the best light-weight CNNs in the tradeoff between accuracy and on-device efficiency.
no code implementations • 2 May 2022 • Xian Shi, Xun Xu, Wanyue Zhang, Xiatian Zhu, Chuan Sheng Foo, Kui Jia
We also demonstrate the feasibility of a more efficient training strategy.
no code implementations • 10 Apr 2022 • Victor Escorcia, Ricardo Guerrero, Xiatian Zhu, Brais Martinez
To overcome both limitations, we introduce Self-Supervised Learning Over Sets (SOS), an approach to pre-train a generic Objects In Contact (OIC) representation model from video object regions detected by an off-the-shelf hand-object contact detector.
no code implementations • 7 Feb 2022 • Mingkun Li, Peng Xu, Xiatian Zhu, Jun Guo
We investigate unsupervised person re-identification (Re-ID) with clothes change, a new challenging problem with more practical usability and scalability to real-world deployment.
1 code implementation • CVPR 2022 • Shuaifeng Li, Mao Ye, Xiatian Zhu, Lihua Zhou, Lin Xiong
This approach suffers from both unsatisfactory accuracy of pseudo labels due to the presence of domain shift and limited use of target domain training data.
no code implementations • NeurIPS 2021 • Mengmeng Xu, Juan Manuel Perez Rua, Xiatian Zhu, Bernard Ghanem, Brais Martinez
This results in a task discrepancy problem for the video encoder – trained for action classification, but used for TAL.
1 code implementation • NeurIPS 2021 • Jiachen Lu, Jinghan Yao, Junge Zhang, Xiatian Zhu, Hang Xu, Weiguo Gao, Chunjing Xu, Tao Xiang, Li Zhang
Crucially, with a linear complexity, much longer token sequences are permitted in SOFT, resulting in superior trade-off between accuracy and complexity.
1 code implementation • 20 Oct 2021 • Sauradip Nag, Xiatian Zhu, Tao Xiang
Further, a novel FS-TAL model is proposed which maximizes the knowledge transfer from training classes whilst enabling the model to be dynamically adapted to both the new class and each video of that class simultaneously.
Ranked #1 on
Few Shot Temporal Action Localization
on ActivityNet
Action Segmentation
Few Shot Temporal Action Localization
+4
no code implementations • 29 Sep 2021 • Sauradip Nag, Xiatian Zhu, Yi-Zhe Song, Tao Xiang
In this paper, to address the above two challenges, a novel {\em Global Segmentation Mask Transformer} (GSMT) is proposed.
no code implementations • 21 Sep 2021 • Feng Zhang, Xiatian Zhu, Chen Wang
Human pose estimation in unconstrained images and videos is a fundamental computer vision task.
no code implementations • 19 Sep 2021 • Chen Wang, Feng Zhang, Xiatian Zhu, Shuzhi Sam Ge
Human pose estimation has achieved significant progress on images with high imaging resolution.
1 code implementation • ICCV 2021 • Zhihe Lu, Sen He, Xiatian Zhu, Li Zhang, Yi-Zhe Song, Tao Xiang
A few-shot semantic segmentation model is typically composed of a CNN encoder, a CNN decoder and a simple classifier (separating foreground and background pixels).
1 code implementation • 28 Jul 2021 • Xiangtai Li, Li Zhang, Guangliang Cheng, Kuiyuan Yang, Yunhai Tong, Xiatian Zhu, Tao Xiang
Modelling long-range contextual relationships is critical for pixel-wise prediction tasks such as semantic segmentation.
1 code implementation • 31 May 2021 • Peng Xu, Xiatian Zhu
Currently, one of the most significant limitations in this field is the lack of a large realistic benchmark.
no code implementations • 28 Mar 2021 • Mengmeng Xu, Juan-Manuel Perez-Rua, Xiatian Zhu, Bernard Ghanem, Brais Martinez
This results in a task discrepancy problem for the video encoder -- trained for action classification, but used for TAL.
no code implementations • 25 Jan 2021 • Mantun Chen, Yongjun Wang, Zhiquan Qin, Xiatian Zhu
To address this, we introduce a model-agnostic, efficient, and Harmonious Data Augmentation (HDA) method that can improve deep WF attacking methods significantly.
Data Augmentation
Cryptography and Security
68M25
K.4.1
1 code implementation • 20 Jan 2021 • Xiatian Zhu, Antoine Toisoul, Juan-Manuel Perez-Rua, Li Zhang, Brais Martinez, Tao Xiang
Extensive experiments on four standard few-shot action benchmarks show that our method clearly outperforms previous state-of-the-art methods, with the improvement particularly significant (10+\%) on the most challenging fine-grained action recognition benchmark.
no code implementations • 16 Jan 2021 • Minxian Li, Xiatian Zhu, Shaogang Gong
Extensive comparative experiments demonstrate that the proposed STL model surpasses significantly the state-of-the-art unsupervised learning and one-shot learning re-id methods on three large tracklet person re-id benchmarks.
5 code implementations • CVPR 2021 • Sixiao Zheng, Jiachen Lu, Hengshuang Zhao, Xiatian Zhu, Zekun Luo, Yabiao Wang, Yanwei Fu, Jianfeng Feng, Tao Xiang, Philip H. S. Torr, Li Zhang
In this paper, we aim to provide an alternative perspective by treating semantic segmentation as a sequence-to-sequence prediction task.
Ranked #1 on
Semantic Segmentation
on FoodSeg103
(using extra training data)
2 code implementations • 8 Dec 2020 • Hui Tang, Xiatian Zhu, Ke Chen, Kui Jia, C. L. Philip Chen
To address this issue, we are motivated by a UDA assumption of structural similarity across domains, and propose to directly uncover the intrinsic target discrimination via constrained clustering, where we constrain the clustering solutions using structural source regularization that hinges on the very same assumption.
1 code implementation • ICCV 2021 • Mengmeng Xu, Juan-Manuel Perez-Rua, Victor Escorcia, Brais Martinez, Xiatian Zhu, Li Zhang, Bernard Ghanem, Tao Xiang
However, most existing models developed for these tasks are pre-trained on general video action classification tasks.
Ranked #13 on
Temporal Action Localization
on ActivityNet-1.3
no code implementations • 3 Jul 2020 • Juan-Manuel Perez-Rua, Antoine Toisoul, Brais Martinez, Victor Escorcia, Li Zhang, Xiatian Zhu, Tao Xiang
In this challenge, action recognition is posed as the problem of simultaneously predicting a single `verb' and `noun' class label given an input trimmed video clip.
no code implementations • 2 Apr 2020 • Juan-Manuel Perez-Rua, Brais Martinez, Xiatian Zhu, Antoine Toisoul, Victor Escorcia, Tao Xiang
Departing from existing alternatives, our W3 module models all three facets of video attention jointly.
Ranked #1 on
Action Recognition
on EgoGesture
no code implementations • 13 Mar 2020 • Hanbin Dai, Liangbo Zhou, Feng Zhang, Zhengyu Zhang, Hong Hu, Xiatian Zhu, Mao Ye
Taking them together, we formulate a novel Distribution-Aware coordinate Representation for Keypoint (DARK) method.
no code implementations • CVPR 2020 • Juan-Manuel Perez-Rua, Xiatian Zhu, Timothy Hospedales, Tao Xiang
To this end we propose OpeN-ended Centre nEt (ONCE), a detector designed for incrementally learning to detect novel class objects with few examples.
no code implementations • 12 Feb 2020 • Xiangping Zhu, Xiatian Zhu, Minxian Li, Pietro Morerio, Vittorio Murino, Shaogang Gong
Existing person re-identification (re-id) methods mostly exploit a large set of cross-camera identity labelled training data.
no code implementations • 30 Dec 2019 • Zhiyi Cheng, Xiatian Zhu, Shaogang Gong
Extensive evaluations demonstrate the performance superiority of our method over state-of-the-art SR and UDA models on both genuine and artificial LR facial imagery data.
6 code implementations • CVPR 2020 • Feng Zhang, Xiatian Zhu, Hanbin Dai, Mao Ye, Ce Zhu
Interestingly, we found that the process of decoding the predicted heatmaps into the final joint coordinates in the original image space is surprisingly significant for human pose estimation performance, which nevertheless was not recognised before.
Ranked #3 on
Multi-Person Pose Estimation
on COCO
(using extra training data)
no code implementations • 25 Sep 2019 • Wei Li, Shaogang Gong, Xiatian Zhu
We address this limitation by additionally exploiting feature self-calibration operations, resulting in a heterogeneous search space.
no code implementations • 27 Aug 2019 • Xiangping Zhu, Xiatian Zhu, Minxian Li, Vittorio Murino, Shaogang Gong
Existing person re-identification (re-id) methods rely mostly on a large set of inter-camera identity labelled training data, requiring a tedious data collection and annotation process therefore leading to poor scalability in practical re-id applications.
no code implementations • 22 Jul 2019 • Xu Lan, Xiatian Zhu, Shaogang Gong
Most state-of-the-art person re-identification (re-id) methods depend on supervised model learning with a large set of cross-view identity labelled training data.
1 code implementation • 25 Apr 2019 • Jiabo Huang, Qi Dong, Shaogang Gong, Xiatian Zhu
Deep convolutional neural networks (CNNs) have demonstrated remarkable success in computer vision by supervisedly learning strong visual feature representations.
1 code implementation • 1 Mar 2019 • Minxian Li, Xiatian Zhu, Shaogang Gong
We formulate an Unsupervised Tracklet Association Learning (UTAL) framework.
Ranked #1 on
Person Re-Identification
on DukeTracklet
no code implementations • 21 Nov 2018 • Zhiyi Cheng, Xiatian Zhu, Shaogang Gong
Whilst recent face-recognition (FR) techniques have made significant progress on recognising constrained high-resolution web images, the same cannot be said on natively unconstrained low-resolution images at large scales.
no code implementations • 20 Nov 2018 • Qi Dong, Xiatian Zhu, Shaogang Gong
The objective learning formulation is essential for the success of convolutional neural networks.
no code implementations • 19 Nov 2018 • Xu Lan, Xiatian Zhu, Shaogang Gong
Whilst being able to create stronger target networks compared to the vanilla non-teacher based learning strategy, this scheme needs to train additionally a large teacher model with expensive computational cost.
1 code implementation • CVPR 2019 • Feng Zhang, Xiatian Zhu, Mao Ye
In this work, we investigate the under-studied but practically critical pose model efficiency problem.
Ranked #9 on
Pose Estimation
on Leeds Sports Poses
no code implementations • 25 Sep 2018 • Aytaç Kanacı, Xiatian Zhu, Shaogang Gong
Existing vehicle re-identification (re-id) evaluation benchmarks consider strongly artificial test scenarios by assuming the availability of high quality images and fine-grained appearance at an almost constant image scale, reminiscent to images required for Automatic Number Plate Recognition, e. g. VeRi-776.
no code implementations • ECCV 2018 • Minxian Li, Xiatian Zhu, Shaogang Gong
Mostexistingpersonre-identification(re-id)methods relyon supervised model learning on per-camera-pair manually labelled pairwise training data.
Ranked #2 on
Person Re-Identification
on DukeTracklet
1 code implementation • ECCV 2018 • Yanbei Chen, Xiatian Zhu, Shaogang Gong
We consider the semi-supervised multi-class classification problem of learning from sparse labelled and abundant unlabelled training data.
1 code implementation • 22 Aug 2018 • Yanbei Chen, Xiatian Zhu, Shaogang Gong
In this work, to address the video person re-id task, we formulate a novel Deep Association Learning (DAL) scheme, the first end-to-end deep learning method using none of the identity labels in model initialisation and training.
Ranked #5 on
Person Re-Identification
on PRID2011
Unsupervised Person Re-Identification
Video-Based Person Re-Identification
no code implementations • ECCV 2018 • Xu Lan, Xiatian Zhu, Shaogang Gong
In contrast to previous studies, we show that sufficiently reliable person instance cropping is achievable by slightly improved state-of-the-art deep learning object detectors (e. g. Faster-RCNN), and the under-studied multi-scale matching problem in person search is a more severe barrier.
2 code implementations • 5 Jul 2018 • Hang Su, Xiatian Zhu, Shaogang Gong
In this work, we introduce a more realistic and challenging logo detection setting, called Open Logo Detection.
no code implementations • 25 Jun 2018 • Hanxiao Wang, Xiatian Zhu, Shaogang Gong, Tao Xiang
Most existing person re-identification (re-id) methods are unsuitable for real-world deployment due to two reasons: Unscalability to large population size, and Inadaptability over time.
3 code implementations • NeurIPS 2018 • Xu Lan, Xiatian Zhu, Shaogang Gong
Knowledge distillation is effective to train small and generalisable network models for meeting the low-memory and fast running requirements.
1 code implementation • 28 Apr 2018 • Qi Dong, Shaogang Gong, Xiatian Zhu
In particular, existing deep learning methods consider mostly either class balanced data or moderately imbalanced data in model training, and ignore the challenge of learning from significantly imbalanced training data.
1 code implementation • 25 Apr 2018 • Zhiyi Cheng, Xiatian Zhu, Shaogang Gong
To facilitate more studies on developing FR models that are effective and robust for low-resolution surveillance facial images, we introduce a new Surveillance Face Recognition Challenge, which we call the QMUL-SurvFace benchmark.
2 code implementations • 30 Mar 2018 • Hang Su, Shaogang Gong, Xiatian Zhu
Existing logo detection methods usually consider a small number of logo classes and limited images per class with a strong assumption of requiring tedious object bounding box annotations, therefore not scalable to real-world dynamic applications.
no code implementations • CVPR 2018 • Jingya Wang, Xiatian Zhu, Shaogang Gong, Wei Li
Most existing person re-identification (re-id) methods require supervised model learning from a separate large set of pairwise labelled training data for every single camera pair.
Ranked #21 on
Unsupervised Domain Adaptation
on Market to Duke
Unsupervised Domain Adaptation
Unsupervised Person Re-Identification
1 code implementation • CVPR 2018 • Wei Li, Xiatian Zhu, Shaogang Gong
Existing person re-identification (re-id) methods either assume the availability of well-aligned person bounding box images as model input or rely on constrained attention selection mechanisms to calibrate misaligned images.
Ranked #12 on
Person Re-Identification
on CUHK03
1 code implementation • ICCV 2017 • Qi Dong, Shaogang Gong, Xiatian Zhu
Recognising detailed facial or clothing attributes in images of people is a challenging task for computer vision, especially when the training data are both in very large scale and extremely imbalanced among different attribute classes.
no code implementations • ICCV 2017 • Jingya Wang, Xiatian Zhu, Shaogang Gong, Wei Li
Recognising semantic pedestrian attributes in surveillance images is a challenging task for computer vision, particularly when the imaging quality is poor with complex background clutter and uncontrolled viewing conditions, and the number of labelled training data is small.
Multi-Label Image Classification
Pedestrian Attribute Recognition
no code implementations • 10 Jul 2017 • Xu Lan, Hanxiao Wang, Shaogang Gong, Xiatian Zhu
Existing person re-identification (re-id) methods assume the provision of accurately cropped person bounding boxes with minimum background noise, mostly by manually cropping.
no code implementations • 30 May 2017 • Jingya Wang, Xiatian Zhu, Shaogang Gong
As a result, our model is able to discover more accurate semantic correlation between textual tags and visual features, and finally providing favourable visual semantics interpretation even with highly sparse and incomplete tags.
no code implementations • 12 May 2017 • Wei Li, Xiatian Zhu, Shaogang Gong
Existing person re-identification (re-id) methods rely mostly on either localised or global feature representation alone.
Ranked #86 on
Person Re-Identification
on Market-1501
no code implementations • 26 Mar 2017 • Ying-Cong Chen, Xiatian Zhu, Wei-Shi Zheng, Jian-Huang Lai
The challenge of person re-identification (re-id) is to match individual images of the same person captured by different non-overlapping camera views against significant and unknown cross-view feature distortion.
no code implementations • 29 Dec 2016 • Hang Su, Xiatian Zhu, Shaogang Gong
Logo detection in unconstrained images is challenging, particularly when only very sparse labelled training images are accessible due to high labelling costs.
no code implementations • 5 Dec 2016 • Hanxiao Wang, Shaogang Gong, Xiatian Zhu, Tao Xiang
Current person re-identification (re-id) methods assume that (1) pre-labelled training data are available for every camera pair, (2) the gallery size for re-identification is moderate.
no code implementations • 25 Nov 2016 • Xiaolong Ma, Xiatian Zhu, Shaogang Gong, Xudong Xie, Jianming Hu, Kin-Man Lam, Yisheng Zhong
Crucially, this model does not require pairwise labelled training data (i. e. unsupervised) therefore readily scalable to large scale camera networks of arbitrary camera pairs without the need for exhaustive data annotation for every camera pair.
no code implementations • 12 Oct 2016 • Qi Dong, Shaogang Gong, Xiatian Zhu
Recognising detailed clothing characteristics (fine-grained attributes) in unconstrained images of people in-the-wild is a challenging task for computer vision, especially when there is only limited training data from the wild whilst most data available for model learning are captured in well-controlled environments using fashion models (well lit, no background clutter, frontal view, high-resolution).
no code implementations • 23 Jan 2016 • Taiqing Wang, Shaogang Gong, Xiatian Zhu, Shengjin Wang
Current person re-identification (ReID) methods typically rely on single-frame imagery features, whilst ignoring space-time information from image sequences often available in the practical surveillance scenarios.
no code implementations • 13 Jan 2015 • Xiatian Zhu, Chen Change Loy, Shaogang Gong
Many visual surveillance tasks, e. g. video summarisation, is conventionally accomplished through analysing imagerybased features.
no code implementations • CVPR 2014 • Xiatian Zhu, Chen Change Loy, Shaogang Gong
Spectral clustering requires robust and meaningful affinity graphs as input in order to form clusters with desired structures that can well support human intuition.