1 code implementation • 16 Mar 2024 • Shichao Kan, Yuhai Deng, Yixiong Liang, Lihui Cen, Zhe Qu, Yigang Cen, Zhihai He
This paper presents a novel unsupervised deep metric learning approach, termed unsupervised collaborative metric learning with mixed-scale groups (MS-UGCML), devised to learn embeddings for objects of varying scales.
no code implementations • 15 Jan 2024 • Yi Zhang, Ce Zhang, Ke Yu, Yushun Tang, Zhihai He
However, for generalization tasks, the current fine-tuning methods for CLIP, such as CoOp and CoCoOp, demonstrate relatively low performance on some fine-grained datasets.
no code implementations • 2 Nov 2023 • Xueting Hu, Ce Zhang, Yi Zhang, Bowen Hai, Ke Yu, Zhihai He
When CLIP is used for depth estimation tasks, the patches, divided from the input images, can be combined with a series of semantic descriptions of the depth information to obtain similarity results.
no code implementations • 3 Sep 2023 • Yi Zhang, Ce Zhang, Zihan Liao, Yushun Tang, Zhihai He
Large-scale pre-trained Vision-Language Models (VLMs), such as CLIP and ALIGN, have introduced a new paradigm for learning transferable visual representations.
no code implementations • 22 Aug 2023 • Yi Zhang, Ce Zhang, Xueting Hu, Zhihai He
To leverage the valuable knowledge encoded within these models for downstream tasks, several fine-tuning approaches, including prompt tuning methods and adapter-based methods, have been developed to adapt vision-language models effectively with supervision.
no code implementations • 28 Jul 2023 • Yi Zhang, Ce Zhang, Yushun Tang, Zhihai He
Based on these visual concepts, we construct a discriminative representation of images and learn a concept inference network to perform downstream image classification tasks, such as few-shot learning and domain generalization.
no code implementations • 29 Jun 2023 • Yushun Tang, Qinghai Guo, Zhihai He
Our main idea is that, when we adapt the network model to predict the sample labels from encoded features, we use these prediction results to construct new training samples with derived labels to learn a new examiner network that performs a different but compatible task in the target domain.
no code implementations • 15 Apr 2023 • Ce Zhang, Kailiang Wu, Zhihai He
Given an unknown dynamical system, what is the minimum number of samples needed for effective learning of its governing laws and accurate prediction of its future evolution behavior, and how to select these critical samples?
no code implementations • CVPR 2023 • Zhehan Kan, Shuoshuo Chen, Ce Zhang, Yushun Tang, Zhihai He
This strong correlation suggests that we can use this error as feedback to guide the correction process.
no code implementations • CVPR 2023 • Yushun Tang, Ce Zhang, Heng Xu, Shuoshuo Chen, Jie Cheng, Luziwei Leng, Qinghai Guo, Zhihai He
We observe that the performance of this feed-forward Hebbian learning for fully test-time adaptation can be significantly improved by incorporating a feedback neuro-modulation layer.
1 code implementation • 10 Oct 2022 • Shichao Kan, Zhiquan He, Yigang Cen, Yang Li, Vladimir Mladenovic, Zhihai He
Recent methods for deep metric learning have been focusing on designing different contrastive loss functions between positive and negative pairs of samples so that the learned feature embedding is able to pull positive samples of the same class closer and push negative samples from different classes away from each other.
no code implementations • 9 Oct 2022 • Shichao Kan, Yixiong Liang, Min Li, Yigang Cen, Jianxin Wang, Zhihai He
To address this challenge, in this paper, we introduce a new method called coded residual transform (CRT) for deep metric learning to significantly improve its generalization capability.
no code implementations • 6 Jul 2022 • Zhehan Kan, Shuoshuo Chen, Zeng Li, Zhihai He
This group-wise structural correlation can be explored to improve the accuracy and robustness of human pose estimation.
Ranked #1 on Multi-Person Pose Estimation on COCO test-dev
no code implementations • 29 Sep 2021 • Wenming Cao, Qifan Liu, Guang Liu, Zhihai He
We construct a prime-dual network structure for few-shot learning which establishes a commutative relationship between the support set and the query set, as well as a new self- supervision constraint for highly effective few-shot learning.
no code implementations • 29 Sep 2021 • Wenming Cao, Zhineng Zhao, Qifan Liu, Zhihai He
Few-shot learning (FSL) aims to characterize the inherent visual relationship between support and query samples which can be well generalized to unseen classes so that we can accurately infer the labels of query samples from very few support samples.
no code implementations • CVPR 2021 • Shichao Kan, Yigang Cen, Yang Li, Vladimir Mladenovic, Zhihai He
During training, this relative order prediction network and the feature embedding network are tightly coupled, providing mutual constraints to each other to improve overall metric learning performance in a cooperative manner.
no code implementations • CVPR 2021 • Yang Li, Shichao Kan, Jianhe Yuan, Wenming Cao, Zhihai He
It has been long recognized that deep neural networks are sensitive to changes in spatial configurations or scene structures.
no code implementations • 4 Mar 2021 • Zhiqun Zhao, Hengyou Wang, Hao Sun, Zhihai He
In this work, we propose to develop a structure-preserving progressive low-rank image completion (SPLIC) method to remove unneeded texture details from the input images and shift the bias of deep neural networks towards global object structures and semantic cues.
no code implementations • ICCV 2021 • Jianhe Yuan, Zhihai He
Specifically, we esti-mate the victim model in the black box using a learned lin-ear composition of an ensemble of surrogate models withdiversified network structures.
no code implementations • ECCV 2020 • Yang Li, Shichao Kan, Zhihai He
To further enhance the inter-class discriminative power of the feature generated by this network, we adapt the concept of triplet loss from supervised metric learning to our unsupervised case and introduce the contrastive clustering loss.
no code implementations • CVPR 2020 • Jianhe Yuan, Zhihai He
In this paper, we develop a new method called ensemble generative cleaning with feedback loops (EGC-FL) for effective defense of deep neural networks.
no code implementations • CVPR 2020 • Hao Sun, Zhiqun Zhao, Zhihai He
Based on this unique property, we develop a new approach, called reciprocal learning, for human trajectory prediction.
no code implementations • 25 Sep 2019 • Jianhe Yuan, Zhihai He
In this paper, we develop a new generative cleaning network with quantized nonlinear transform for effective defense of deep neural networks.
no code implementations • 4 Sep 2019 • Yang Li, Jianhe Yuan, Zhiqun Zhao, Hao Sun, Zhihai He
In this work, we develop a joint sample discovery and iterative model evolution method for semi-supervised learning on very small labeled training sets.
no code implementations • 25 Apr 2018 • Zhi Zhang, Guanghan Ning, Yigang Cen, Yang Li, Zhiqun Zhao, Hao Sun, Zhihai He
The inference structures and computational complexity of existing deep neural networks, once trained, are fixed and remain the same for all test images.
no code implementations • 27 Oct 2017 • Guanghan Ning, Zhihai He
The task of multi-person human pose estimation in natural scenes is quite challenging.
no code implementations • 26 Oct 2017 • Zhi Zhang, Guanghan Ning, Zhihai He
In this paper, we will develop a new framework for training deep neural networks on datasets with limited labeled samples using cross-network knowledge projection which is able to improve the network performance while reducing the overall computational complexity significantly.
1 code implementation • 18 Jul 2017 • Chiara Zizza, Adam Starr, Devin Hudson, Sai Shreya Nuguri, Prasad Calyam, Zhihai He
Virtual Learning Environments (VLEs) are spaces designed to educate students remotely via online platforms.
Human-Computer Interaction Multimedia
1 code implementation • 5 May 2017 • Guanghan Ning, Zhi Zhang, Zhihai He
Human pose estimation using deep neural networks aims to map input images with large variations into multiple body keypoints which must satisfy a set of geometric constraints and inter-dependency imposed by the human body model.
Ranked #6 on Pose Estimation on Leeds Sports Poses
2 code implementations • 19 Jul 2016 • Guanghan Ning, Zhi Zhang, Chen Huang, Zhihai He, Xiaobo Ren, Haohong Wang
In this paper, we develop a new approach of spatially supervised recurrent convolutional neural networks for visual object tracking.
no code implementations • 7 Apr 2016 • Miao Sun, Tony X. Han, Zhihai He
Currently, the state-of-the-art image classification algorithms outperform the best available object detector by a big margin in terms of average precision.
no code implementations • CVPR 2013 • Xiaobo Ren, Tony X. Han, Zhihai He
We incorporate this similarity information into a graph-cut energy minimization framework for foreground object segmentation.