1 code implementation • 3 Apr 2024 • Huayi Zhou, Fei Jiang, Hongtao Lu
Existing head pose estimation datasets are either composed of numerous samples by non-realistic synthesis or lab collection, or limited images by labor-intensive annotating.
no code implementations • 28 Mar 2024 • Weihao Jiang, Zhaozhi Xie, Yuxiang Lu, Longjie Qi, Jingyong Cai, Hiroyuki Uchiyama, Bin Chen, Yue Ding, Hongtao Lu
Our framework and model introduce the following key aspects: (1) to learn real-world adaptive semantic representation for objects with diverse and complex structures under real-world scenes, we introduce extra semantic segmentation and edge detection tasks on more diverse real-world data with segmentation annotations; (2) to avoid overfitting on low-level details, we propose a module to utilize the inconsistency between learned segmentation and matting representations to regularize detail refinement; (3) we propose a novel background line detection task into our auxiliary learning framework, to suppress interference of background lines or textures.
no code implementations • 16 Mar 2024 • Deyi Ji, Siqi Gao, Lanyun Zhu, Yiru Zhao, Peng Xu, Hongtao Lu, Feng Zhao
In this paper, we address the challenge of multi-object tracking (MOT) in moving Unmanned Aerial Vehicle (UAV) scenarios, where irregular flight trajectories, such as hovering, turning left/right, and moving up/down, lead to significantly greater complexity compared to fixed-camera MOT.
no code implementations • 1 Mar 2024 • Suizhi Huang, Shalayiding Sirejiding, Yuxiang Lu, Yue Ding, Leheng Liu, Hui Zhou, Hongtao Lu
Object detection and semantic segmentation are pivotal components in biomedical image analysis.
no code implementations • 1 Mar 2024 • Yuxiang Lu, Shalayiding Sirejiding, Bayram Bayramli, Suizhi Huang, Yue Ding, Hongtao Lu
The task-conditional model is a distinctive stream for efficient multi-task learning.
1 code implementation • 20 Feb 2024 • Yuwen Yang, Yuxiang Lu, Suizhi Huang, Shalayiding Sirejiding, Hongtao Lu, Yue Ding
The innovative Federated Multi-Task Learning (FMTL) approach consolidates the benefits of Federated Learning (FL) and Multi-Task Learning (MTL), enabling collaborative model training on multi-task learning datasets.
1 code implementation • 18 Feb 2024 • Huayi Zhou, Mukun Luo, Fei Jiang, Yue Ding, Hongtao Lu
The 2D human pose estimation (HPE) is a basic visual problem.
no code implementations • 24 Jan 2024 • Ziru Zeng, Yue Ding, Hongtao Lu
Recently, the detection transformer has gained substantial attention for its inherent minimal post-processing requirement. However, this paradigm relies on abundant training data, yet in the context of the cross-domain adaptation, insufficient labels in the target domain exacerbate issues of class imbalance and model performance degradation. To address these challenges, we propose a novel class-aware cross domain detection transformer based on the adversarial learning and mean-teacher framework. First, considering the inconsistencies between the classification and regression tasks, we introduce an IoU-aware prediction branch and exploit the consistency of classification and location scores to filter and reweight pseudo labels. Second, we devise a dynamic category threshold refinement to adaptively manage model confidence. Third, to alleviate the class imbalance, an instance-level class-aware contrastive learning module is presented to encourage the generation of discriminative features for each class, particularly benefiting minority classes. Experimental results across diverse domain-adaptive scenarios validate our method's effectiveness in improving performance and alleviating class imbalance issues, which outperforms the state-of-the-art transformer based methods.
1 code implementation • 23 Jan 2024 • Zhaozhi Xie, Bochen Guan, Weihao Jiang, Muyang Yi, Yue Ding, Hongtao Lu, Lei Zhang
In this paper, we introduce a novel prompt-driven adapter into SAM, namely Prompt Adapter Segment Anything Model (PA-SAM), aiming to enhance the segmentation mask quality of the original SAM.
no code implementations • 29 Dec 2023 • Deyi Ji, Siqi Gao, Mingyuan Tao, Hongtao Lu, Feng Zhao
The ChangeNet dataset is suitable for both binary change detection (BCD) and semantic change detection (SCD) tasks.
no code implementations • 22 Nov 2023 • Yuxiang Lu, Suizhi Huang, Yuwen Yang, Shalayiding Sirejiding, Yue Ding, Hongtao Lu
Moreover, we employ learnable Hyper Aggregation Weights for each client to customize personalized parameter updates.
no code implementations • 15 Nov 2023 • Xiaoshuang Chen, Zhongyi Sun, Ke Yan, Shouhong Ding, Hongtao Lu
In detail, CPPF consists of a prototype clustering module (PC), an embedding space reserving module (ESR) and a multi-teacher distillation module (MTD).
no code implementations • 16 Sep 2023 • Yuwen Yang, Chang Liu, Xun Cai, Suizhi Huang, Hongtao Lu, Yue Ding
Federated Learning (FL) has emerged as a promising approach to enable collaborative learning among multiple clients while preserving data privacy.
no code implementations • 28 Jul 2023 • Yuxiang Lu, Shalayiding Sirejiding, Yue Ding, Chunlin Wang, Hongtao Lu
Task-conditional architecture offers advantage in parameter efficiency but falls short in performance compared to state-of-the-art multi-decoder methods.
no code implementations • 3 Jul 2023 • Deyi Ji, Feng Zhao, Hongtao Lu
For the sake of high inference speed and low computation complexity, $\mathcal{T}$ partitions the original UHR image into patches and groups them dynamically, then learns the low-level local details with the lightweight multi-head Wavelet Transformer (WFormer) network.
1 code implementation • 15 Jun 2023 • Tianyu Li, Subhankar Roy, Huayi Zhou, Hongtao Lu, Stephane Lathuiliere
To address this, we present CONtrastive FEaTure and pIxel alignment (CONFETI) for bridging the domain gap at both the pixel and feature levels using a unique contrastive formulation.
1 code implementation • CVPR 2023 • Deyi Ji, Feng Zhao, Hongtao Lu, Mingyuan Tao, Jieping Ye
With the increasing interest and rapid development of methods for Ultra-High Resolution (UHR) segmentation, a large-scale benchmark covering a wide range of scenes with full fine-grained dense annotations is urgently needed to facilitate the field.
Ranked #1 on Semantic Segmentation on INRIA Aerial Image Labeling (mIOU metric)
no code implementations • CVPR 2022 • Deyi Ji, Haoran Wang, Mingyuan Tao, Jianqiang Huang, Xian-Sheng Hua, Hongtao Lu
Existing knowledge distillation works for semantic segmentation mainly focus on transferring high-level contextual knowledge from teacher to student.
1 code implementation • 21 Apr 2023 • Huayi Zhou, Fei Jiang, Jiaxin Si, Yue Ding, Hongtao Lu
In this paper, we focus on the joint detection of human body and its parts.
1 code implementation • 2 Feb 2023 • Huayi Zhou, Fei Jiang, Hongtao Lu
We present comprehensive comparisons with state-of-the-art single HPE methods on public benchmarks, as well as superior baseline results on our constructed MPHPE datasets.
1 code implementation • 15 Jan 2023 • Jianrong Zhang, Yangsong Zhang, Xiaodong Cun, Shaoli Huang, Yong Zhang, Hongwei Zhao, Hongtao Lu, Xi Shen
Additionally, we conduct analyses on HumanML3D and observe that the dataset size is a limitation of our approach.
Ranked #11 on Motion Synthesis on HumanML3D
no code implementations • CVPR 2023 • Jianrong Zhang, Yangsong Zhang, Xiaodong Cun, Yong Zhang, Hongwei Zhao, Hongtao Lu, Xi Shen, Ying Shan
Additionally, we conduct analyses on HumanML3D and observe that the dataset size is a limitation of our approach.
1 code implementation • CVPR 2023 • Muyang Yi, Quan Cui, Hao Wu, Cheng Yang, Osamu Yoshie, Hongtao Lu
LoDA and SimSeg jointly ameliorate a vanilla CLIP to produce impressive semantic segmentation results.
1 code implementation • 15 Dec 2022 • Huayi Zhou, Fei Jiang, Hongtao Lu
This paper focuses on the problem of joint detection of human body and its corresponding parts.
no code implementations • 7 Dec 2022 • Huayi Zhou, Fei Jiang, Lili Xiong, Hongtao Lu
Most recent head pose estimation (HPE) methods are dominated by the Euler angle representation.
Ranked #8 on Head Pose Estimation on BIWI (MAE (trained with BIWI data) metric)
no code implementations • 19 Nov 2022 • Chang Liu, Yuwen Yang, Yue Ding, Hongtao Lu
While most existing message-passing graph neural networks (MPNNs) are permutation-invariant in graph-level representation learning and permutation-equivariant in node- and edge-level representation learning, their expressive power is commonly limited by the 1-Weisfeiler-Lehman (1-WL) graph isomorphism test.
1 code implementation • 6 Nov 2022 • Huayi Zhou, Fei Jiang, Jiaxin Si, Lili Xiong, Hongtao Lu
In this paper, we present StuArt, a novel automatic system designed for the individualized classroom observation, which empowers instructors to concern the learning status of each student.
1 code implementation • 4 Nov 2022 • Huayi Zhou, Fei Jiang, Hongtao Lu
Domain adaptive object detection (DAOD) aims to alleviate transfer performance degradation caused by the cross-domain discrepancy.
1 code implementation • 1 Nov 2022 • Chang Liu, Yuwen Yang, Zhe Xie, Hongtao Lu, Yue Ding
2) Prevailing graph augmentation methods for GEL, including rule-based, sample-based, adaptive, and automated methods, are not suitable for augmenting subgraphs because a subgraph contains fewer nodes but richer information such as position, neighbor, and structure.
no code implementations • 28 Oct 2022 • Chang Liu, Yuwen Yang, Xun Cai, Yue Ding, Hongtao Lu
Federated learning (FL) faces three major difficulties: cross-domain, heterogeneous models, and non-i. i. d.
1 code implementation • 27 Oct 2022 • Huayi Zhou, Fei Jiang, Jiaxin Si, Hongtao Lu
In the paper, we propose a single-stage end-to-end trainable framework for tackling the HBOE problem with multi-persons.
no code implementations • 13 Oct 2022 • Chang Liu, Yuwen Yang, Yue Ding, Hongtao Lu
The normalizing layer has become one of the basic configurations of deep learning models, but it still suffers from computational inefficiency, interpretability difficulties, and low generality.
1 code implementation • 4 Oct 2022 • Yangsong Zhang, Subhankar Roy, Hongtao Lu, Elisa Ricci, Stéphane Lathuilière
In this work we address multi-target domain adaptation (MTDA) in semantic segmentation, which consists in adapting a single model from an annotated source dataset to multiple unannotated target datasets that differ in their underlying data distributions.
no code implementations • 3 Jul 2022 • Fuzhi Yang, Huan Yang, Yanhong Zeng, Jianlong Fu, Hongtao Lu
The extractor estimates the degradations in LR inputs and guides the meta-restoration modules to predict restoration parameters for different degradations on-the-fly.
no code implementations • 3 May 2022 • Bayram Bayramli, Junhwa Hur, Hongtao Lu
Self-supervised methods demonstrate learning scene flow estimation from unlabeled data, yet their accuracy lags behind (semi-)supervised methods.
no code implementations • 22 Feb 2022 • Xiaoshuang Chen, Hongtao Lu
To address this problem, we propose a self-adaptive feature similarity learning (SFSL) network and a global-local consistency (GLC) loss to reinforce local feature representation.
1 code implementation • 19 Feb 2022 • Huayi Zhou, Fei Jiang, Hongtao Lu
Video surveillance systems have been installed to ensure the student safety in schools.
1 code implementation • 1 Dec 2021 • Weihao Jiang, Dongdong Yu, Zhaozhi Xie, Yaoyi Li, Zehuan Yuan, Hongtao Lu
For emerging content-based feature fusion, most existing matting methods only focus on local features which lack the guidance of a global feature with strong semantic information related to the interesting object.
Ranked #4 on Image Matting on Composition-1K
no code implementations • 21 Nov 2021 • Runyuan Cai, Yue Ding, Hongtao Lu
A specialized pipeline is designed, and we further propose a frequency loss function to fit the nature of our frequency-domain task.
no code implementations • 31 Oct 2021 • Xiaoshuang Chen, Yiru Zhao, Yu Qin, Fei Jiang, Mingyuan Tao, Xiansheng Hua, Hongtao Lu
Crowd counting aims to learn the crowd density distributions and estimate the number of objects (e. g. persons) in images.
1 code implementation • 19 Mar 2021 • Zhe Xie, Chengxuan Liu, Yichi Zhang, Hongtao Lu, Dong Wang, Yue Ding
To solve the above problem, in this work, we propose a novel method called Adversarial and Contrastive Variational Autoencoder (ACVAE) for sequential recommendation.
no code implementations • 1 Sep 2020 • Usman Ali, Bayram Bayramli, Hongtao Lu
Most existing person re-id methods generally require a large amount of identity labeled data to act as discriminative guideline for representation learning.
1 code implementation • CVPR 2020 • Fuzhi Yang, Huan Yang, Jianlong Fu, Hongtao Lu, Baining Guo
In this paper, we propose a novel Texture Transformer Network for Image Super-Resolution (TTSR), in which the LR and Ref images are formulated as queries and keys in a transformer, respectively.
no code implementations • 4 Jun 2020 • Weihao Jiang, Zhaozhi Xie, Yaoyi Li, Chang Liu, Hongtao Lu
Many of these applications need to perform a real-time and efficient prediction for semantic segmentation with a light-weighted network.
no code implementations • 7 Apr 2020 • Yaoyi Li, Qingyao Xu, Hongtao Lu
Natural image matting is a fundamental problem in computational photography and computer vision.
no code implementations • 15 Jan 2020 • Li Wang, Zechen Bai, Yonghua Zhang, Hongtao Lu
SG and RWS are de-signed for the best use of recalled words.
1 code implementation • 13 Jan 2020 • Yaoyi Li, Hongtao Lu
Inspired by affinity-based method and the successes of contextual attention in inpainting, we develop a novel end-to-end approach for natural image matting with a guided contextual attention module, which is specifically designed for image matting.
no code implementations • 16 May 2019 • Yaoyi Li, Jianfu Zhang, Weijie Zhao, Hongtao Lu
A high efficient image matting method based on a weakly annotated mask is in demand for mobile applications.
no code implementations • 16 May 2019 • Bayram Bayramli, Usman Ali, Te Qi, Hongtao Lu
One of the most important problem is low resolution face images which can result in bad performance on face recognition.
no code implementations • 5 Apr 2019 • Te Qi, Bayram Bayramli, Usman Ali, Qinchuan Zhang, Hongtao Lu
Like many computer vision problems, human pose estimation is a challenging problem in that recognizing a body part requires not only information from local area but also from areas with large spatial distance.
7 code implementations • 1 Jan 2019 • Wenbo Li, Zhicheng Wang, Binyi Yin, Qixiang Peng, Yuming Du, Tianzi Xiao, Gang Yu, Hongtao Lu, Yichen Wei, Jian Sun
Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods.
Ranked #1 on Pose Estimation on COCO minival
no code implementations • 8 Dec 2018 • Jiajun Du, Yu Qin, Hongtao Lu, Yonghua Zhang
Most attention-based image captioning models attend to the image once per word.
no code implementations • ECCV 2018 • Yiru Zhao, Zhongming Jin, Guo-Jun Qi, Hongtao Lu, Xian-Sheng Hua
While deep neural networks have demonstrated competitive results for many visual recognition and image retrieval tasks, the major challenge lies in distinguishing similar images from different categories (i. e., hard negative examples) while clustering images with large variations from the same category (i. e., hard positive examples).
no code implementations • 2 Dec 2017 • Zihao Hu, Xiyi Luo, Hongtao Lu, Yong Yu
Recently, supervised hashing methods have attracted much attention since they can optimize retrieval speed and storage cost while preserving semantic information.
no code implementations • CVPR 2017 • Zihao Hu, Junxuan Chen, Hongtao Lu, Tongzhen Zhang
To address this problem, we present a novel fully Bayesian treatment for supervised hashing problem, named Bayesian Supervised Hashing (BSH), in which hyperparameters are automatically tuned during optimization.
no code implementations • 28 Mar 2017 • Fei Jiang, Xiao-Yang Liu, Hongtao Lu, Ruimin Shen
Sparse coding (SC) is an automatic feature extraction and selection technique that is widely used in unsupervised learning.
no code implementations • 27 Mar 2017 • Fei Jiang, Xiao-Yang Liu, Hongtao Lu, Ruimin Shen
Sparse coding (SC) is an unsupervised learning scheme that has received an increasing amount of interests in recent years.
no code implementations • 30 Oct 2016 • Shicong Liu, Hongtao Lu
However, how to learn deep representations that strongly preserve similarities between data pairs and can be accurately quantized via vector quantization remains a challenging task.
no code implementations • 21 Sep 2016 • Yaoyi Li, Hongtao Lu
In this paper, we present a method dubbed Consensus Prior Constraint Propagation (CPCP), which can provide the prior knowledge of the robustness of each data instance and its neighborhood.
no code implementations • 20 Sep 2016 • Junxuan Chen, Baigui Sun, Hao Li, Hongtao Lu, Xian-Sheng Hua
Click through rate (CTR) prediction of image ads is the core task of online display advertising systems, and logistic regression (LR) has been frequently applied as the prediction model.
no code implementations • 13 Nov 2015 • Yaoyi Li, Junxuan Chen, Hongtao Lu
In this paper, we propose a novel method, dubbed Adaptive Affinity Matrix (AdaAM), to learn an adaptive affinity matrix and derive a distance metric from the affinity.
no code implementations • 28 Sep 2015 • Xuan Luo, Xuejiao Bai, Shuo Li, Hongtao Lu, Sei-ichiro Kamata
This is partially because our DTs overcome the extreme greediness of the MST.
no code implementations • 17 Sep 2015 • Shicong Liu, Junru Shao, Hongtao Lu
Further, we propose Aggregating-Tree (A-Tree), a non-exhaustive search method using HCLAE to perform efficient ANN-Search.
no code implementations • 17 Sep 2015 • Shicong Liu, Junru Shao, Hongtao Lu
We propose a novel distance to calculate distance between high dimensional vector pairs, utilizing vector quantization generated encodings.
no code implementations • 17 Sep 2015 • Shicong Liu, Hongtao Lu, Junru Shao
In this paper, we propose an improved residual vector quantization (IRVQ) method, our IRVQ learns codebook with a hybrid method of subspace clustering and warm-started k-means on each stage to prevent performance gain from dropping, and uses a multi-path encoding scheme to encode a vector with lower distortion.
no code implementations • 6 Jul 2015 • Shicong Liu, Hongtao Lu
Jointly used with residual vector quantization, our optimized dictionaries lead to a better approximate nearest neighbor search performance compared to the state-of-the-art methods.