no code implementations • 22 Mar 2024 • Heng Guo, Jianfeng Zhang, Jiaxing Huang, Tony C. W. Mok, Dazhou Guo, Ke Yan, Le Lu, Dakai Jin, Minfeng Xu
In this work, we propose a comprehensive and scalable 3D SAM model for whole-body CT segmentation, named CT-SAM3D.
1 code implementation • 12 Mar 2024 • Han Qiu, Jiaxing Huang, Peng Gao, Lewei Lu, Xiaoqin Zhang, Shijian Lu
Inspired by the success of general-purpose models in NLP, recent studies attempt to unify different vision tasks in the same sequence format and employ autoregressive Transformers for sequence prediction.
no code implementations • 7 Feb 2024 • Sheng Jin, Xueying Jiang, Jiaxing Huang, Lewei Lu, Shijian Lu
This paper presents DVDet, a Descriptor-Enhanced Open Vocabulary Detector that introduces conditional context prompts and hierarchical textual descriptors that enable precise region-text alignment as well as open-vocabulary detection training in general.
no code implementations • 13 Jan 2024 • Kai Jiang, Jiaxing Huang, Weiying Xie, Yunsong Li, Ling Shao, Shijian Lu
Large-vocabulary object detectors (LVDs) aim to detect objects of many categories, which learn super objectness features and can locate objects accurately while applied to various downstream data.
no code implementations • 13 Jan 2024 • Kai Jiang, Jiaxing Huang, Weiying Xie, Yunsong Li, Ling Shao, Shijian Lu
Camera-only Bird's Eye View (BEV) has demonstrated great potential in environment perception in a 3D space.
no code implementations • 9 Jan 2024 • Jiaxing Huang, Kai Jiang, Jingyi Zhang, Han Qiu, Lewei Lu, Shijian Lu, Eric Xing
SAMs work with two types of prompts including spatial prompts (e. g., points) and semantic prompts (e. g., texts), which work together to prompt SAMs to segment anything on downstream datasets.
no code implementations • 27 Dec 2023 • Jiaxing Huang, Jingyi Zhang, Kai Jiang, Han Qiu, Shijian Lu
Traditional computer vision generally solves each single task independently by a dedicated model with the task instruction implicitly designed in the model architecture, arising two limitations: (1) it leads to task-specific models, which require multiple models for different tasks and restrict the potential synergies from diverse tasks; (2) it leads to a pre-defined and fixed model interface that has limited interactivity and adaptability in following user' task instructions.
no code implementations • ICCV 2023 • Xueying Jiang, Jiaxing Huang, Sheng Jin, Shijian Lu
Despite its recent progress, most existing work suffers from the misalignment between the difficulty level of training samples and the capability of contemporarily trained models, leading to over-fitting or under-fitting in the trained generalization model.
no code implementations • ICCV 2023 • Jingyi Zhang, Jiaxing Huang, Xueying Jiang, Shijian Lu
However, the source predictions of target data are often noisy and training with them is prone to learning collapses.
no code implementations • 29 Jun 2023 • Jiaxing Huang, Jingyi Zhang, Han Qiu, Sheng Jin, Shijian Lu
Traditional domain adaptation assumes the same vocabulary across source and target domains, which often struggles with limited transfer flexibility and efficiency while handling target domains with different vocabularies.
1 code implementation • CVPR 2023 • Aoran Xiao, Jiaxing Huang, Weihao Xuan, Ruijie Ren, Kangcheng Liu, Dayan Guan, Abdulmotaleb El Saddik, Shijian Lu, Eric Xing
In addition, we design a domain randomization technique that alternatively randomizes the geometry styles of point clouds and aggregates their embeddings, ultimately leading to a generalizable model that can improve 3DSS under various adverse weather effectively.
1 code implementation • 3 Apr 2023 • Jingyi Zhang, Jiaxing Huang, Sheng Jin, Shijian Lu
Most visual recognition studies rely heavily on crowd-labelled data in deep neural networks (DNNs) training, and they usually train a DNN for each single visual recognition task, leading to a laborious and time-consuming visual recognition paradigm.
1 code implementation • ICCV 2023 • Yanfeng Zhou, Jiaxing Huang, Chenlong Wang, Le Song, Ge Yang
Perturbations in consistency-based semi-supervised models are often artificially designed.
2 code implementations • 30 Jul 2022 • Aoran Xiao, Jiaxing Huang, Dayan Guan, Kaiwen Cui, Shijian Lu, Ling Shao
The first is scene-level swapping which exchanges point cloud sectors of two LiDAR scans that are cut along the azimuth axis.
1 code implementation • 28 Jul 2022 • Gongjie Zhang, Zhipeng Luo, Jiaxing Huang, Shijian Lu, Eric P. Xing
The recently proposed DEtection TRansformer (DETR) has established a fully end-to-end paradigm for object detection.
no code implementations • 26 Jul 2022 • Chuhui Xue, Jiaxing Huang, Shijian Lu, Changhu Wang, Song Bai
We formulate the new setup by a dual detection task which first detects integral text units and then groups them into a CTB.
1 code implementation • 6 Jul 2022 • Yun Xing, Dayan Guan, Jiaxing Huang, Shijian Lu
Specifically, we design cross-frame pseudo labelling to provide pseudo supervision from previous video frames while learning from the augmented current video frames.
no code implementations • CVPR 2023 • Jingyi Zhang, Jiaxing Huang, Xiaoqin Zhang, Shijian Lu
Domain adaptive panoptic segmentation aims to mitigate data annotation challenge by leveraging off-the-shelf annotated data in one or multiple related source domains.
Ranked #2 on Domain Adaptation on Panoptic SYNTHIA-to-Cityscapes
1 code implementation • CVPR 2022 • Dayan Guan, Jiaxing Huang, Aoran Xiao, Shijian Lu
We build the balanced subclass distributions by clustering pixels of each original class into multiple subclasses of similar sizes, which provide class-balanced pseudo supervision to regularize the class-biased segmentation.
1 code implementation • 28 Feb 2022 • Aoran Xiao, Jiaxing Huang, Dayan Guan, Xiaoqin Zhang, Shijian Lu, Ling Shao
The convergence of point cloud and DNNs has led to many deep point cloud models, largely trained under the supervision of large-scale and densely-labelled point cloud data.
1 code implementation • NeurIPS 2021 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu
To this end, we design an innovative historical contrastive learning (HCL) technique that exploits historical source hypothesis to make up for the absence of source data in UMA.
1 code implementation • 4 Oct 2021 • Kaiwen Cui, Jiaxing Huang, Zhipeng Luo, Gongjie Zhang, Fangneng Zhan, Shijian Lu
Specifically, we design GenCo, a Generative Co-training network that mitigates the discriminator over-fitting issue by introducing multiple complementary discriminators that provide diverse supervision from multiple distinctive views in training.
no code implementations • 29 Sep 2021 • Chuhui Xue, Jiaxing Huang, Wenqing Zhang, Shijian Lu, Song Bai, Changhu Wang
This paper presents Contextual Text Detection, a new setup that detects contextual text blocks for better understanding of texts in scenes.
1 code implementation • ICCV 2021 • Dayan Guan, Jiaxing Huang, Aoran Xiao, Shijian Lu
This paper presents DA-VSN, a domain adaptive video segmentation network that addresses domain gaps in videos by temporal consistency regularization (TCR) for consecutive frames of target-domain videos.
1 code implementation • 12 Jul 2021 • Aoran Xiao, Jiaxing Huang, Dayan Guan, Fangneng Zhan, Shijian Lu
Extensive experiments show that SynLiDAR provides a high-quality data source for studying 3D transfer and the proposed PCT achieves superior point cloud translation consistently across the three setups.
no code implementations • 7 Jul 2021 • Kaiwen Cui, Gongjie Zhang, Fangneng Zhan, Jiaxing Huang, Shijian Lu
Generative Adversarial Networks (GANs) have become the de-facto standard in image synthesis.
no code implementations • CVPR 2022 • Jingyi Zhang, Jiaxing Huang, Zichen Tian, Shijian Lu
Second, it introduces multi-view spectral learning that learns useful unsupervised representations by maximizing mutual information among multiple ST-generated spectral views of each target sample.
no code implementations • 5 Jun 2021 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu
We position the few labeled target samples as references that gauge the similarity between source and target features and guide adaptive inter-domain alignment for learning more similar source features.
1 code implementation • ICCV 2021 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu
With FAA-generated samples, the training can continue the 'random walk' and drift into an area with a flat loss landscape, leading to more robust domain adaptation.
1 code implementation • CVPR 2022 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu, Ling Shao
In this work, we explore the idea of instance contrastive learning in unsupervised domain adaptation (UDA) and propose a novel Category Contrast technique (CaCo) that introduces semantic priors on top of instance discrimination for visual UDA tasks.
no code implementations • 18 May 2021 • Chuhui Xue, Jiaxing Huang, Wenqing Zhang, Shijian Lu, Changhu Wang, Song Bai
The first task focuses on image-to-character (I2C) mapping which detects a set of character candidates from images based on different alignments of visual features in an non-sequential way.
no code implementations • CVPR 2023 • Jingyi Zhang, Jiaxing Huang, Zhipeng Luo, Gongjie Zhang, Xiaoqin Zhang, Shijian Lu
DA-DETR introduces a novel CNN-Transformer Blender (CTBlender) that fuses the CNN features and Transformer features ingeniously for effective feature alignment and knowledge transfer across domains.
no code implementations • 24 Mar 2021 • Jiaxing Huang, Dayan Guan, Shijian Lu, Aoran Xiao
Recent progresses in domain adaptive semantic segmentation demonstrate the effectiveness of adversarial learning (AL) in unsupervised domain adaptation.
1 code implementation • CVPR 2021 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu
It has been studied widely by domain randomization that transfers source images to different styles in spatial space for learning domain-agnostic features.
1 code implementation • CVPR 2021 • Jiaxing Huang, Dayan Guan, Aoran Xiao, Shijian Lu
The inter-task regularization exploits the complementary nature of instance segmentation and semantic segmentation and uses it as a constraint for better feature alignment across domains.
Ranked #2 on Domain Adaptation on Panoptic SYNTHIA-to-Mapillary
1 code implementation • 1 Mar 2021 • Aoran Xiao, Xiaofei Yang, Shijian Lu, Dayan Guan, Jiaxing Huang
Specifically, we design a residual dense block with multiple receptive fields as a building block in the encoder which preserves detailed information in each modality and learns hierarchical modality-specific and fused features effectively.
Ranked #23 on 3D Semantic Segmentation on SemanticKITTI
3 code implementations • 27 Feb 2021 • Dayan Guan, Jiaxing Huang, Aoran Xiao, Shijian Lu, Yanpeng Cao
Specifically, we design an uncertainty metric that assesses the alignment of each sample and adjusts the strength of adversarial learning for well-aligned and poorly-aligned samples adaptively.
1 code implementation • ECCV 2020 • Jiaxing Huang, Shijian Lu, Dayan Guan, Xiaobing Zhang
Recent advances in unsupervised domain adaptation for semantic segmentation have shown great potentials to relieve the demand of expensive per-pixel annotations.
no code implementations • 12 May 2019 • Fangneng Zhan, Jiaxing Huang, Shijian Lu
Despite the rapid progress of generative adversarial networks (GANs) in image synthesis in recent years, the existing image synthesis approaches work in either geometry domain or appearance domain alone which often introduces various synthesis artifacts.