1 code implementation • ECCV 2020 • Xiao Zhang, Rui Zhao, Yu Qiao, Hongsheng Li
To address this problem, this paper introduces a novel Radial Basis Function (RBF) distances to replace the commonly used inner products in the softmax loss function, such that it can adaptively assign losses to regularize the intra-class and inter-class distances by reshaping the relative differences, and thus creating more representative prototypes of classes to improve optimization.
no code implementations • 27 Mar 2024 • Sicheng Li, Keqiang Sun, Zhixin Lai, Xiaoshi Wu, Feng Qiu, Haoran Xie, Kazunori Miyata, Hongsheng Li
Secondly, to overcome the issue of limited conditional supervision, we introduce Diffusion Consistency Loss (DCL), which applies supervision on the denoised latent code at any given time step.
1 code implementation • 25 Mar 2024 • Hao Shao, Shengju Qian, Han Xiao, Guanglu Song, Zhuofan Zong, Letian Wang, Yu Liu, Hongsheng Li
This paper presents Visual CoT, a novel pipeline that leverages the reasoning capabilities of multi-modal large language models (MLLMs) by incorporating visual Chain-of-Thought (CoT) reasoning.
no code implementations • 21 Mar 2024 • Renrui Zhang, Dongzhi Jiang, Yichi Zhang, Haokun Lin, Ziyu Guo, Pengshuo Qiu, Aojun Zhou, Pan Lu, Kai-Wei Chang, Peng Gao, Hongsheng Li
To this end, we introduce MathVerse, an all-around visual math benchmark designed for an equitable and in-depth evaluation of MLLMs.
1 code implementation • 20 Mar 2024 • Fu-Yun Wang, Xiaoshi Wu, Zhaoyang Huang, Xiaoyu Shi, Dazhong Shen, Guanglu Song, Yu Liu, Hongsheng Li
We introduce MOTIA Mastering Video Outpainting Through Input-Specific Adaptation, a diffusion-based pipeline that leverages both the intrinsic data-specific patterns of the source video and the image/video generative prior for effective outpainting.
1 code implementation • 19 Mar 2024 • Linjiang Huang, Rongyao Fang, Aiping Zhang, Guanglu Song, Si Liu, Yu Liu, Hongsheng Li
In this study, we delve into the generation of high-resolution images from pre-trained diffusion models, addressing persistent challenges, such as repetitive patterns and structural distortions, that emerge when models are applied beyond their trained resolutions.
1 code implementation • 18 Mar 2024 • Yang Zhou, Hao Shao, Letian Wang, Steven L. Waslander, Hongsheng Li, Yu Liu
Context information, such as road maps and surrounding agents' states, provides crucial geometric and semantic information for motion behavior prediction.
2 code implementations • 14 Mar 2024 • Haiyang Wang, Hao Tang, Li Jiang, Shaoshuai Shi, Muhammad Ferjad Naeem, Hongsheng Li, Bernt Schiele, LiWei Wang
Due to its simple design, this paradigm holds promise for narrowing the architectural gap between vision and language.
1 code implementation • 4 Mar 2024 • Yuchen Duan, Weiyun Wang, Zhe Chen, Xizhou Zhu, Lewei Lu, Tong Lu, Yu Qiao, Hongsheng Li, Jifeng Dai, Wenhai Wang
Our evaluations demonstrate that VRWKV surpasses ViT's performance in image classification and has significantly faster speeds and lower memory usage processing high-resolution inputs.
no code implementations • 26 Feb 2024 • Zimu Lu, Aojun Zhou, Houxing Ren, Ke Wang, Weikang Shi, Junting Pan, Mingjie Zhan, Hongsheng Li
We augment the ground-truth solutions of our seed data and train a back-translation model to translate the augmented solutions back into new questions.
no code implementations • 22 Feb 2024 • Ke Wang, Junting Pan, Weikang Shi, Zimu Lu, Mingjie Zhan, Hongsheng Li
Recent advancements in Large Multimodal Models (LMMs) have shown promising results in mathematical reasoning within visual contexts, with models approaching human-level performance on existing benchmarks such as MathVista.
Ranked #1 on Multimodal Reasoning on MATH-V (using extra training data)
1 code implementation • 22 Feb 2024 • Xudong Lu, Qi Liu, Yuhui Xu, Aojun Zhou, Siyuan Huang, Bo Zhang, Junchi Yan, Hongsheng Li
Specifically, we propose, for the first time to our best knowledge, post-training approaches for task-agnostic and task-specific expert pruning and skipping of MoE LLMs, tailored to improve deployment efficiency while maintaining model performance across a wide range of tasks.
1 code implementation • 8 Feb 2024 • Peng Gao, Renrui Zhang, Chris Liu, Longtian Qiu, Siyuan Huang, Weifeng Lin, Shitian Zhao, Shijie Geng, Ziyi Lin, Peng Jin, Kaipeng Zhang, Wenqi Shao, Chao Xu, Conghui He, Junjun He, Hao Shao, Pan Lu, Hongsheng Li, Yu Qiao
We propose SPHINX-X, an extensive Multimodality Large Language Model (MLLM) series developed upon SPHINX.
Ranked #16 on Visual Question Answering on MM-Vet
1 code implementation • 1 Feb 2024 • Fu-Yun Wang, Zhaoyang Huang, Xiaoyu Shi, Weikang Bian, Guanglu Song, Yu Liu, Hongsheng Li
We validate the proposed strategy in image-conditioned video generation and layout-conditioned video generation, all achieving top-performing results.
no code implementations • 29 Jan 2024 • Xiaoyu Shi, Zhaoyang Huang, Fu-Yun Wang, Weikang Bian, Dasong Li, Yi Zhang, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, Hongsheng Li
For the first stage, we propose a diffusion-based motion field predictor, which focuses on deducing the trajectories of the reference image's pixels.
1 code implementation • 18 Jan 2024 • Changyao Tian, Xizhou Zhu, Yuwen Xiong, Weiyun Wang, Zhe Chen, Wenhai Wang, Yuntao Chen, Lewei Lu, Tong Lu, Jie zhou, Hongsheng Li, Yu Qiao, Jifeng Dai
Developing generative models for interleaved image-text data has both research and practical value.
no code implementations • 13 Jan 2024 • Jingqiu Zhou, Aojun Zhou, Hongsheng Li
Moreover, we studied the robustness of OOD methods by applying different types of image encoders.
Out-of-Distribution Detection Out of Distribution (OOD) Detection
1 code implementation • 11 Jan 2024 • Yuwen Xiong, Zhiqi Li, Yuntao Chen, Feng Wang, Xizhou Zhu, Jiapeng Luo, Wenhai Wang, Tong Lu, Hongsheng Li, Yu Qiao, Lewei Lu, Jie zhou, Jifeng Dai
The advancements in speed and efficiency of DCNv4, combined with its robust performance across diverse vision tasks, show its potential as a foundational building block for future vision models.
no code implementations • 10 Jan 2024 • Zhanliang He, Nuoye Xiong, Hongsheng Li, Peiyi Shen, Guangming Zhu, Liang Zhang
Through experimental validation, based on this interaction interface, NN can provide humans with easily understandable explanations of the reasoning process.
1 code implementation • 8 Jan 2024 • Huanyu Liu, JianFeng Cai, Tingjia Zhang, Hongsheng Li, Siyuan Wang, Guangming Zhu, Syed Afaq Ali Shah, Mohammed Bennamoun, Liang Zhang
Automated conversion methods are essential to overcome manual conversion challenges.
no code implementations • 21 Dec 2023 • Keqiang Sun, Dor Litvak, Yunzhi Zhang, Hongsheng Li, Jiajun Wu, Shangzhe Wu
We introduce Ponymation, a new method for learning a generative model of articulated 3D animal motions from raw, unlabeled online videos.
2 code implementations • 19 Dec 2023 • Chaoyou Fu, Renrui Zhang, Zihan Wang, Yubo Huang, Zhengye Zhang, Longtian Qiu, Gaoxiang Ye, Yunhang Shen, Mengdan Zhang, Peixian Chen, Sirui Zhao, Shaohui Lin, Deqiang Jiang, Di Yin, Peng Gao, Ke Li, Hongsheng Li, Xing Sun
They endow Large Language Models (LLMs) with powerful capabilities in visual understanding, enabling them to tackle diverse multi-modal tasks.
no code implementations • 14 Dec 2023 • Hao Li, Xue Yang, Zhaokai Wang, Xizhou Zhu, Jie zhou, Yu Qiao, Xiaogang Wang, Hongsheng Li, Lewei Lu, Jifeng Dai
Traditional reinforcement-learning-based agents rely on sparse rewards that often only use binary values to indicate task completion or failure.
1 code implementation • 12 Dec 2023 • Hao Shao, Yuxuan Hu, Letian Wang, Steven L. Waslander, Yu Liu, Hongsheng Li
On the other hand, previous autonomous driving methods tend to rely on limited-format inputs (e. g. sensor data and navigation waypoints), restricting the vehicle's ability to understand language information and interact with humans.
1 code implementation • 30 Nov 2023 • Rongyao Fang, Shilin Yan, Zhaoyang Huang, Jingqiu Zhou, Hao Tian, Jifeng Dai, Hongsheng Li
In this work, we introduce InstructSeq, an instruction-conditioned multi-modal modeling framework that unifies diverse vision tasks through flexible natural language control and handling of both visual and textual data.
1 code implementation • 21 Nov 2023 • Xiaoyu Yang, Lijian Xu, Hongsheng Li, Shaoting Zhang
This approach enables us to optimally utilize the knowledge and reasoning capacities of large pre-trained language models for an array of tasks encompassing both language and vision.
1 code implementation • 13 Nov 2023 • Ziyi Lin, Chris Liu, Renrui Zhang, Peng Gao, Longtian Qiu, Han Xiao, Han Qiu, Chen Lin, Wenqi Shao, Keqin Chen, Jiaming Han, Siyuan Huang, Yichi Zhang, Xuming He, Hongsheng Li, Yu Qiao
We present SPHINX, a versatile multi-modal large language model (MLLM) with a joint mixing of model weights, tuning tasks, and visual embeddings.
Ranked #2 on Visual Question Answering on BenchLMM
1 code implementation • 2 Nov 2023 • Lijian Xu, Ziyu Ni, Xinglong Liu, Xiaosong Wang, Hongsheng Li, Shaoting Zhang
We first compose a multi-task training dataset comprising 13. 4 million instruction and ground-truth pairs (with approximately one million radiographs) for the customized tuning, involving both image- and pixel-level tasks.
no code implementations • 25 Oct 2023 • Manyuan Zhang, Bingqi Ma, Guanglu Song, Yunxiao Wang, Hongsheng Li, Yu Liu
During the COVID-19 coronavirus epidemic, almost everyone is wearing masks, which poses a huge challenge for deep learning-based face recognition algorithms.
no code implementations • ICCV 2023 • Manyuan Zhang, Guanglu Song, Yu Liu, Hongsheng Li
We observe that different regions of interest in the visual feature map are suitable for performing query classification and box localization tasks, even for the same object.
1 code implementation • NeurIPS 2023 • Yazhe Niu, Yuan Pu, Zhenjie Yang, Xueyan Li, Tong Zhou, Jiyuan Ren, Shuai Hu, Hongsheng Li, Yu Liu
Building agents based on tree-search planning capabilities with learned models has achieved remarkable success in classic decision-making problems, such as Go and Atari.
no code implementations • 12 Oct 2023 • Hu Yu, Li Shen, Jie Huang, Man Zhou, Hongsheng Li, Feng Zhao
Diffusion models have demonstrated compelling generation quality by optimizing the variational lower bound through a simple denoising score matching loss.
1 code implementation • 5 Oct 2023 • Ke Wang, Houxing Ren, Aojun Zhou, Zimu Lu, Sichun Luo, Weikang Shi, Renrui Zhang, Linqi Song, Mingjie Zhan, Hongsheng Li
In this paper, we present a method to fine-tune open-source language models, enabling them to use code for modeling and deriving math equations and, consequently, enhancing their mathematical reasoning abilities.
Ranked #4 on Math Word Problem Solving on SVAMP (using extra training data)
1 code implementation • ICCV 2023 • Jiawei Yao, Chuming Li, Keqiang Sun, Yingjie Cai, Hao Li, Wanli Ouyang, Hongsheng Li
Monocular 3D Semantic Scene Completion (SSC) has garnered significant attention in recent years due to its potential to predict complex semantics and geometry shapes from a single image, requiring no 3D inputs.
3D Semantic Scene Completion from a single 2D image 3D Semantic Scene Completion from a single RGB image
2 code implementations • 7 Sep 2023 • Jiaming Han, Renrui Zhang, Wenqi Shao, Peng Gao, Peng Xu, Han Xiao, Kaipeng Zhang, Chris Liu, Song Wen, Ziyu Guo, Xudong Lu, Shuai Ren, Yafei Wen, Xiaoxin Chen, Xiangyu Yue, Hongsheng Li, Yu Qiao
During training, we adopt a learnable bind network to align the embedding space between LLaMA and ImageBind's image encoder.
5 code implementations • 1 Sep 2023 • Ziyu Guo, Renrui Zhang, Xiangyang Zhu, Yiwen Tang, Xianzheng Ma, Jiaming Han, Kexin Chen, Peng Gao, Xianzhi Li, Hongsheng Li, Pheng-Ann Heng
We introduce Point-Bind, a 3D multi-modality model aligning point clouds with 2D image, language, audio, and video.
Ranked #5 on 3D Question Answering (3D-QA) on 3D MM-Vet
no code implementations • ICCV 2023 • Jinyu Chen, Wenguan Wang, Si Liu, Hongsheng Li, Yi Yang
CCPD transfers the fundamental, point-to-point wayfinding skill that is well trained on the large-scale PointGoal task to ORAN, so as to help ORAN to better master audio-visual navigation with far fewer training samples.
1 code implementation • 15 Aug 2023 • Aojun Zhou, Ke Wang, Zimu Lu, Weikang Shi, Sichun Luo, Zipeng Qin, Shaoqing Lu, Anya Jia, Linqi Song, Mingjie Zhan, Hongsheng Li
We found that its success can be largely attributed to its powerful skills in generating and executing code, evaluating the output of code execution, and rectifying its solution when receiving unreasonable outputs.
Ranked #1 on Math Word Problem Solving on MATH
1 code implementation • 7 Aug 2023 • Wenqi Shao, Yutao Hu, Peng Gao, Meng Lei, Kaipeng Zhang, Fanqing Meng, Peng Xu, Siyuan Huang, Hongsheng Li, Yu Qiao, Ping Luo
Secondly, it conducts an in-depth analysis of LVLMs' predictions using the ChatGPT Ensemble Evaluation (CEE), which leads to a robust and accurate evaluation and exhibits improved alignment with human evaluation compared to the word matching approach.
1 code implementation • 20 Jul 2023 • Yiyuan Zhang, Kaixiong Gong, Kaipeng Zhang, Hongsheng Li, Yu Qiao, Wanli Ouyang, Xiangyu Yue
Multimodal learning aims to build models that can process and relate information from multiple modalities.
1 code implementation • ICCV 2023 • Fan Lu, Yan Xu, Guang Chen, Hongsheng Li, Kwan-Yee Lin, Changjun Jiang
To construct urban-level radiance fields efficiently, we design Deformable Neural Mesh Primitive~(DNMP), and propose to parameterize the entire scene with such primitives.
1 code implementation • 15 Jun 2023 • Xiaoshi Wu, Yiming Hao, Keqiang Sun, Yixiong Chen, Feng Zhu, Rui Zhao, Hongsheng Li
By fine-tuning CLIP on HPD v2, we obtain Human Preference Score v2 (HPS v2), a scoring model that can more accurately predict human preferences on generated images.
no code implementations • 15 Jun 2023 • Junting Pan, Ziyi Lin, Yuying Ge, Xiatian Zhu, Renrui Zhang, Yi Wang, Yu Qiao, Hongsheng Li
Video Question Answering (VideoQA) has been significantly advanced from the scaling of recent Large Language Models (LLMs).
Ranked #3 on Temporal/Casual QA on NExT-QA (using extra training data)
1 code implementation • ICCV 2023 • Tao Ma, Xuemeng Yang, Hongbin Zhou, Xin Li, Botian Shi, Junjie Liu, Yuchen Yang, Zhizheng Liu, Liang He, Yu Qiao, Yikang Li, Hongsheng Li
Extensive experiments on Waymo Open Dataset show our DetZero outperforms all state-of-the-art onboard and offboard 3D detection methods.
1 code implementation • ICCV 2023 • Xuesong Chen, Shaoshuai Shi, Chao Zhang, Benjin Zhu, Qiang Wang, Ka Chun Cheung, Simon See, Hongsheng Li
3D multi-object tracking (MOT) is vital for many applications including autonomous driving vehicles and service robots.
no code implementations • 8 Jun 2023 • Zhaoyang Huang, Xiaoyu Shi, Chao Zhang, Qiang Wang, Yijin Li, Hongwei Qin, Jifeng Dai, Xiaogang Wang, Hongsheng Li
This paper introduces a novel transformer-based network architecture, FlowFormer, along with the Masked Cost Volume AutoEncoding (MCVA) for pretraining it to tackle the problem of optical flow estimation.
no code implementations • 8 Jun 2023 • Changyao Tian, Chenxin Tao, Jifeng Dai, Hao Li, Ziheng Li, Lewei Lu, Xiaogang Wang, Hongsheng Li, Gao Huang, Xizhou Zhu
In each denoising step, our method first decodes pixels from previous VQ tokens, then generates new VQ tokens from the decoded pixels.
no code implementations • 3 Jun 2023 • Weikang Bian, Zhaoyang Huang, Xiaoyu Shi, Yitong Dong, Yijin Li, Hongsheng Li
We tackle the problem of Persistent Independent Particles (PIPs), also called Tracking Any Point (TAP), in videos, which specifically aims at estimating persistent long-term trajectories of query points in videos.
no code implementations • 2 Jun 2023 • Zeqiang Lai, Yuchen Duan, Jifeng Dai, Ziheng Li, Ying Fu, Hongsheng Li, Yu Qiao, Wenhai Wang
In this paper, we propose to ameliorate the semantic segmentation quality of existing discriminative approaches with a mask prior modeled by a recently-developed denoising diffusion generative model.
1 code implementation • 1 Jun 2023 • Xiaoliang Ju, Zhaoyang Huang, Yijin Li, Guofeng Zhang, Yu Qiao, Hongsheng Li
In addition to the scene generation, the final part of DiffInDScene can be used as a post-processing module to refine the 3D reconstruction results from multi-view stereo.
no code implementations • 30 May 2023 • Ziyu Ni, Linda Wei, Lijian Xu, Simon Yu, Qing Xia, Hongsheng Li, Shaoting Zhang
In this work, we proposed an end-to-end deep learning framework, which could predict the coronary artery hemodynamics from CCTA images.
1 code implementation • 29 May 2023 • Fu-Yun Wang, Wenshuo Chen, Guanglu Song, Han-Jia Ye, Yu Liu, Hongsheng Li
To address this challenge, we introduce a novel paradigm dubbed as Gen-L-Video, capable of extending off-the-shelf short video diffusion models for generating and editing videos comprising hundreds of frames with diverse semantic segments without introducing additional training, all while preserving content consistency.
1 code implementation • 18 May 2023 • Siyuan Huang, Zhengkai Jiang, Hao Dong, Yu Qiao, Peng Gao, Hongsheng Li
This paper presents Instruct2Act, a framework that utilizes Large Language Models to map multi-modal instructions to sequential actions for robotic manipulation tasks.
no code implementations • CVPR 2023 • Hao Shao, Letian Wang, RuoBing Chen, Steven L. Waslander, Hongsheng Li, Yu Liu
The large-scale deployment of autonomous vehicles is yet to come, and one of the major remaining challenges lies in urban dense traffic scenarios.
Ranked #1 on Autonomous Driving on CARLA Leaderboard
2 code implementations • 16 May 2023 • Siyuan Huang, Bo Zhang, Botian Shi, Peng Gao, Yikang Li, Hongsheng Li
In this paper, different from previous 2D DG works, we focus on the 3D DG problem and propose a Single-dataset Unified Generalization (SUG) framework that only leverages a single source dataset to alleviate the unforeseen domain differences faced by a well-trained source model.
no code implementations • 7 May 2023 • Xiaoyu Yang, Lijian Xu, Simon Yu, Qing Xia, Hongsheng Li, Shaoting Zhang
3) A dataset named CCA-200 is collected, consisting of 200 CCTA images with coronary artery disease.
1 code implementation • 4 May 2023 • Renrui Zhang, Zhengkai Jiang, Ziyu Guo, Shilin Yan, Junting Pan, Xianzheng Ma, Hao Dong, Peng Gao, Hongsheng Li
Driven by large-data pre-training, Segment Anything Model (SAM) has been demonstrated as a powerful and promptable framework, revolutionizing the segmentation models.
Ranked #1 on Personalized Segmentation on PerSeg
3 code implementations • 28 Apr 2023 • Peng Gao, Jiaming Han, Renrui Zhang, Ziyi Lin, Shijie Geng, Aojun Zhou, Wei zhang, Pan Lu, Conghui He, Xiangyu Yue, Hongsheng Li, Yu Qiao
This strategy effectively alleviates the interference between the two tasks of image-text alignment and instruction following and achieves strong multi-modal reasoning with only a small-scale image-text and instruction dataset.
Ranked #6 on Visual Question Answering (VQA) on InfiMM-Eval
Instruction Following Optical Character Recognition (OCR) +7
no code implementations • 19 Apr 2023 • Xiaoliang Ju, Yiyang Sun, Yiming Hao, Yikang Li, Yu Qiao, Hongsheng Li
We propose a perception imitation method to simulate results of a certain perception model, and discuss a new heuristic route of autonomous driving simulator without data synthesis.
1 code implementation • CVPR 2023 • Jingqiu Zhou, Linjiang Huang, Liang Wang, Si Liu, Hongsheng Li
Besides, the generated pseudo-labels can be fluctuating and inaccurate at the early stage of training.
Pseudo Label Weakly-supervised Temporal Action Localization +1
1 code implementation • ICCV 2023 • Zhuofan Zong, Dongzhi Jiang, Guanglu Song, Zeyue Xue, Jingyong Su, Hongsheng Li, Yu Liu
The HoP approach is straightforward: given the current timestamp t, we generate a pseudo Bird's-Eye View (BEV) feature of timestamp t-k from its adjacent frames and utilize this feature to predict the object set at timestamp t-k. Our approach is motivated by the observation that enforcing the detector to capture both the spatial location and temporal motion of objects occurring at historical timestamps can lead to more accurate BEV feature learning.
Ranked #3 on 3D Object Detection on nuScenes Camera Only
7 code implementations • 28 Mar 2023 • Renrui Zhang, Jiaming Han, Chris Liu, Peng Gao, Aojun Zhou, Xiangfei Hu, Shilin Yan, Pan Lu, Hongsheng Li, Yu Qiao
We present LLaMA-Adapter, a lightweight adaption method to efficiently fine-tune LLaMA into an instruction-following model.
Ranked #2 on Music Question Answering on MusicQA
1 code implementation • ICCV 2023 • Xiaoshi Wu, Keqiang Sun, Feng Zhu, Rui Zhao, Hongsheng Li
To address this issue, we collect a dataset of human choices on generated images from the Stable Foundation Discord channel.
1 code implementation • CVPR 2023 • Xiaoshi Wu, Feng Zhu, Rui Zhao, Hongsheng Li
To overcome these obstacles, we propose CORA, a DETR-style framework that adapts CLIP for Open-vocabulary detection by Region prompting and Anchor pre-matching.
1 code implementation • ICCV 2023 • Jihao Liu, Tai Wang, Boxiao Liu, Qihang Zhang, Yu Liu, Hongsheng Li
In this paper, we propose Geometry Enhanced Masked Image Modeling (GeoMIM) to transfer the knowledge of the LiDAR model in a pretrain-finetune paradigm for improving the multi-view camera-based 3D detection.
1 code implementation • ICCV 2023 • Xiaoyu Shi, Zhaoyang Huang, Weikang Bian, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, Hongsheng Li
We first propose a TRi-frame Optical Flow (TROF) module that estimates bi-directional optical flows for the center frame in a three-frame manner.
no code implementations • CVPR 2023 • Junjie Ni, Yijin Li, Zhaoyang Huang, Hongsheng Li, Hujun Bao, Zhaopeng Cui, Guofeng Zhang
However, estimating scale differences between these patches is non-trivial since the scale differences are determined by both relative camera poses and scene structures, and thus spatially varying over image pairs.
2 code implementations • 14 Mar 2023 • Renrui Zhang, Liuhui Wang, Ziyu Guo, Yali Wang, Peng Gao, Hongsheng Li, Jianbo Shi
We present a Non-parametric Network for 3D point cloud analysis, Point-NN, which consists of purely non-learnable components: farthest point sampling (FPS), k-nearest neighbors (k-NN), and pooling operations, with trigonometric functions.
Ranked #1 on Training-free 3D Part Segmentation on ShapeNet-Part
3D Point Cloud Classification Training-free 3D Part Segmentation +1
no code implementations • 14 Mar 2023 • Yijin Li, Zhaoyang Huang, Shuo Chen, Xiaoyu Shi, Hongsheng Li, Hujun Bao, Zhaopeng Cui, Guofeng Zhang
BlinkSim consists of a configurable rendering engine and a flexible engine for event data simulation.
1 code implementation • 9 Mar 2023 • Peng Gao, Renrui Zhang, Rongyao Fang, Ziyi Lin, Hongyang Li, Hongsheng Li, Qiao Yu
To alleviate this, previous methods simply replace the pixel reconstruction targets of 75% masked tokens by encoded features from pre-trained image-image (DINO) or image-language (CLIP) contrastive learning.
1 code implementation • 6 Mar 2023 • Yi Zhang, Dasong Li, Xiaoyu Shi, Dailan He, Kangning Song, Xiaogang Wang, Hongwei Qin, Hongsheng Li
In this paper, we propose a kernel basis attention (KBA) module, which introduces learnable kernel bases to model representative image patterns for spatial information aggregation.
Ranked #1 on Color Image Denoising on McMaster sigma50
3 code implementations • CVPR 2023 • Renrui Zhang, Xiangfei Hu, Bohao Li, Siyuan Huang, Hanqiu Deng, Hongsheng Li, Yu Qiao, Peng Gao
Our CaFo incorporates CLIP's language-contrastive knowledge, DINO's vision-contrastive knowledge, DALL-E's vision-generative knowledge, and GPT-3's language-generative knowledge.
1 code implementation • CVPR 2023 • Xiaoyu Shi, Zhaoyang Huang, Dasong Li, Manyuan Zhang, Ka Chun Cheung, Simon See, Hongwei Qin, Jifeng Dai, Hongsheng Li
FlowFormer introduces a transformer architecture into optical flow estimation and achieves state-of-the-art performance.
1 code implementation • 2 Mar 2023 • Rongyao Fang, Peng Gao, Aojun Zhou, Yingjie Cai, Si Liu, Jifeng Dai, Hongsheng Li
The first method is One-to-many Matching via Data Augmentation (denoted as DataAug-DETR).
1 code implementation • CVPR 2023 • Chen Gao, Xingyu Peng, Mi Yan, He Wang, Lirong Yang, Haibing Ren, Hongsheng Li, Si Liu
In this paper, we propose an Adaptive Zone-aware Hierarchical Planner (AZHP) to explicitly divides the navigation process into two heterogeneous phases, i. e., sub-goal setting via zone partition/selection (high-level action) and sub-goal executing (low-level action), for hierarchical planning.
no code implementations • ICCV 2023 • Aojun Zhou, Yang Li, Zipeng Qin, Jianbo Liu, Junting Pan, Renrui Zhang, Rui Zhao, Peng Gao, Hongsheng Li
In this paper, we aim to reduce model complexity from large vision transformers pretrained by MAE with assistant of sparse training.
1 code implementation • CVPR 2023 • Renrui Zhang, Liuhui Wang, Yali Wang, Peng Gao, Hongsheng Li, Jianbo Shi
We present a Non-parametric Network for 3D point cloud analysis, Point-NN, which consists of purely non-learnable components: farthest point sampling (FPS), k-nearest neighbors (k-NN), and pooling operations, with trigonometric functions.
1 code implementation • CVPR 2023 • Benjin Zhu, Zhe Wang, Shaoshuai Shi, Hang Xu, Lanqing Hong, Hongsheng Li
We thus propose a Query Contrast mechanism to explicitly enhance queries towards their best-matched GTs over all unmatched query predictions.
2 code implementations • CVPR 2023 • Renrui Zhang, Liuhui Wang, Yu Qiao, Peng Gao, Hongsheng Li
Pre-training by numerous image data has become de-facto for robust 2D representations.
Ranked #2 on 3D Point Cloud Linear Classification on ModelNet40 (using extra training data)
3D Point Cloud Linear Classification Few-Shot 3D Point Cloud Classification
no code implementations • 23 Nov 2022 • Keqiang Sun, Shangzhe Wu, Ning Zhang, Zhaoyang Huang, Quan Wang, Hongsheng Li
Capitalizing on the recent advances in image generation models, existing controllable face image synthesis methods are able to generate high-fidelity images with some levels of controllability, e. g., controlling the shapes, expressions, textures, and poses of the generated face images.
1 code implementation • 22 Nov 2022 • Linjiang Huang, Kaixin Lu, Guanglu Song, Liang Wang, Si Liu, Yu Liu, Hongsheng Li
In this paper, we present a novel training scheme, namely Teach-DETR, to learn better DETR-based detectors from versatile teacher detectors.
2 code implementations • CVPR 2023 • Hao Li, Jinguo Zhu, Xiaohu Jiang, Xizhou Zhu, Hongsheng Li, Chun Yuan, Xiaohua Wang, Yu Qiao, Xiaogang Wang, Wenhai Wang, Jifeng Dai
In this paper, we propose Uni-Perceiver v2, which is the first generalist model capable of handling major large-scale vision and vision-language tasks with competitive performance.
2 code implementations • CVPR 2023 • Wenhai Wang, Jifeng Dai, Zhe Chen, Zhenhang Huang, Zhiqi Li, Xizhou Zhu, Xiaowei Hu, Tong Lu, Lewei Lu, Hongsheng Li, Xiaogang Wang, Yu Qiao
Compared to the great progress of large-scale vision transformers (ViTs) in recent years, large-scale models based on convolutional neural networks (CNNs) are still in an early state.
Ranked #1 on Instance Segmentation on COCO test-dev (AP50 metric, using extra training data)
no code implementations • 25 Sep 2022 • Renrui Zhang, Bohao Li, Wei zhang, Hao Dong, Hongsheng Li, Peng Gao, Yu Qiao
In this paper, we propose CoMo, a Collaboration of pre-trained Models that incorporates diverse prior knowledge from various pre-training paradigms for better few-shot learning.
no code implementations • 19 Sep 2022 • Zhaoyang Huang, Xiaokun Pan, Weihong Pan, Weikang Bian, Yan Xu, Ka Chun Cheung, Guofeng Zhang, Hongsheng Li
We tackle the problem of estimating correspondences from a general marker, such as a movie poster, to an image that captures such a marker.
no code implementations • 19 Sep 2022 • Zhe Wang, Hongsheng Li, Qinwei Zhang, Jing Yuan, Xiaogang Wang
Adaptively learning a distance metric from the undersampled training data can significantly improve the matching accuracy of the query fingerprints.
no code implementations • 29 Aug 2022 • Manyuan Zhang, Guanglu Song, Yu Liu, Hongsheng Li
To eliminate the bias of single-aspect research and provide an overall understanding of the face recognition model design, we first carefully design the search space for each aspect, then a comprehensive search method is introduced to jointly search optimal data cleaning, architecture, and loss function design.
1 code implementation • 10 Aug 2022 • Dasong Li, Yi Zhang, Ka Chun Cheung, Xiaogang Wang, Hongwei Qin, Hongsheng Li
With the integration, MSDI-Net can handle various and complicated blurry patterns adaptively.
Ranked #11 on Image Deblurring on GoPro
2 code implementations • 6 Aug 2022 • Ziyi Lin, Shijie Geng, Renrui Zhang, Peng Gao, Gerard de Melo, Xiaogang Wang, Jifeng Dai, Yu Qiao, Hongsheng Li
Video recognition has been dominated by the end-to-end learning paradigm -- first initializing a video recognition model with weights of a pretrained image model and then conducting end-to-end training on videos.
Ranked #26 on Action Classification on Kinetics-400 (using extra training data)
1 code implementation • 28 Jul 2022 • Hao Shao, Letian Wang, RuoBing Chen, Hongsheng Li, Yu Liu
Large-scale deployment of autonomous vehicles has been continually delayed due to safety concerns.
Ranked #2 on Autonomous Driving on CARLA Leaderboard
3 code implementations • 19 Jul 2022 • Renrui Zhang, Zhang Wei, Rongyao Fang, Peng Gao, Kunchang Li, Jifeng Dai, Yu Qiao, Hongsheng Li
On top of that, the performance of Tip-Adapter can be further boosted to be state-of-the-art on ImageNet by fine-tuning the cache model for 10$\times$ fewer epochs than existing methods, which is both effective and efficient.
1 code implementation • 18 Jul 2022 • Jihao Liu, Boxiao Liu, Hang Zhou, Hongsheng Li, Yu Liu
In this paper, we propose a novel data augmentation technique TokenMix to improve the performance of vision transformers.
2 code implementations • 12 Jul 2022 • Jihao Liu, Xin Huang, Guanglu Song, Hongsheng Li, Yu Liu
Finally, we integrate configurable operators and DSMs into a unified search space and search with a Reinforcement Learning-based search algorithm to fully explore the optimal combination of the operators.
Ranked #12 on Neural Architecture Search on ImageNet
1 code implementation • 27 Jun 2022 • Junting Pan, Ziyi Lin, Xiatian Zhu, Jing Shao, Hongsheng Li
This has led to a new research direction in parameter-efficient transfer learning.
Ranked #23 on Action Recognition on Something-Something V2 (using extra training data)
1 code implementation • CVPR 2023 • Dasong Li, Xiaoyu Shi, Yi Zhang, Ka Chun Cheung, Simon See, Xiaogang Wang, Hongwei Qin, Hongsheng Li
In this study, we propose a simple yet effective framework for video restoration.
Ranked #1 on Deblurring on GoPro (using extra training data)
1 code implementation • 19 Jun 2022 • Jiageng Mao, Shaoshuai Shi, Xiaogang Wang, Hongsheng Li
Autonomous driving, in recent years, has been receiving increasing attention for its potential to relieve drivers' burdens and improve the safety of driving.
no code implementations • 16 Jun 2022 • Keqiang Sun, Shangzhe Wu, Zhaoyang Huang, Ning Zhang, Quan Wang, Hongsheng Li
Capitalizing on the recent advances in image generation models, existing controllable face image synthesis methods are able to generate high-fidelity images with some levels of controllability, e. g., controlling the shapes, expressions, textures, and poses of the generated face images.
1 code implementation • 9 Jun 2022 • Jinguo Zhu, Xizhou Zhu, Wenhai Wang, Xiaohua Wang, Hongsheng Li, Xiaogang Wang, Jifeng Dai
To mitigate such interference, we introduce the Conditional Mixture-of-Experts (Conditional MoEs) to generalist models.
no code implementations • 7 Jun 2022 • Hongsheng Li, Guangming Zhu, Wu Zhen, Lan Ni, Peiyi Shen, Liang Zhang, Ning Wang, Cong Hua
However, there is still room for improvement in video HOI detection performance.
3 code implementations • 28 May 2022 • Renrui Zhang, Ziyu Guo, Rongyao Fang, Bin Zhao, Dong Wang, Yu Qiao, Hongsheng Li, Peng Gao
By fine-tuning on downstream tasks, Point-M2AE achieves 86. 43% accuracy on ScanObjectNN, +3. 36% to the second-best, and largely benefits the few-shot classification, part segmentation and 3D object detection with the hierarchical pre-training scheme.
Ranked #4 on 3D Point Cloud Linear Classification on ModelNet40 (using extra training data)
1 code implementation • CVPR 2023 • Jihao Liu, Xin Huang, Jinliang Zheng, Yu Liu, Hongsheng Li
In this paper, we propose Mixed and Masked AutoEncoder (MixMAE), a simple but efficient pretraining method that is applicable to various hierarchical Vision Transformers.
Ranked #2 on Image Classification on Places205
1 code implementation • 12 May 2022 • Xuesong Chen, Shaoshuai Shi, Benjin Zhu, Ka Chun Cheung, Hang Xu, Hongsheng Li
Accurate and reliable 3D detection is vital for many applications including autonomous driving vehicles and service robots.
no code implementations • 10 May 2022 • Dasong Li, Yi Zhang, Ka Lung Law, Xiaogang Wang, Hongwei Qin, Hongsheng Li
As for each sub-network, we propose an efficient multi-frequency denoising network to remove noise of different frequencies.
4 code implementations • 8 May 2022 • Peng Gao, Teli Ma, Hongsheng Li, Ziyi Lin, Jifeng Dai, Yu Qiao
Masked auto-encoding for feature pretraining and multi-scale hybrid convolution-transformer architectures can further unleash the potentials of ViT, leading to state-of-the-art performances on image classification, detection and semantic segmentation.
1 code implementation • 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.
1 code implementation • 25 Apr 2022 • Wei Cheng, Su Xu, Jingtan Piao, Chen Qian, Wayne Wu, Kwan-Yee Lin, Hongsheng Li
Specifically, we compress the light fields for novel view human rendering as conditional implicit neural radiance fields from both geometry and appearance aspects.
1 code implementation • ICCV 2023 • Siming Fan, Jingtan Piao, Chen Qian, Kwan-Yee Lin, Hongsheng Li
In this work, we tackle the problem of real-world fluid animation from a still image.
1 code implementation • CVPR 2022 • Haiyang Wang, Shaoshuai Shi, Ze Yang, Rongyao Fang, Qi Qian, Hongsheng Li, Bernt Schiele, LiWei Wang
In order to learn better representations of object shape to enhance cluster features for predicting 3D boxes, we propose a ray-based feature grouping module, which aggregates the point-wise features on object surfaces using a group of determined rays uniformly emitted from cluster centers.
Ranked #12 on 3D Object Detection on ScanNetV2
1 code implementation • 30 Mar 2022 • Zhaoyang Huang, Xiaoyu Shi, Chao Zhang, Qiang Wang, Ka Chun Cheung, Hongwei Qin, Jifeng Dai, Hongsheng Li
We introduce optical Flow transFormer, dubbed as FlowFormer, a transformer-based neural network architecture for learning optical flow.
Ranked #1 on Optical Flow Estimation on Sintel-final
no code implementations • CVPR 2022 • Yingjie Cai, Kwan-Yee Lin, Chao Zhang, Qiang Wang, Xiaogang Wang, Hongsheng Li
Specifically, we map a series of related partial point clouds into multiple complete shape and occlusion code pairs and fuse the codes to obtain their representations in the unified latent space.
1 code implementation • CVPR 2022 • Yan Xu, Kwan-Yee Lin, Guofeng Zhang, Xiaogang Wang, Hongsheng Li
The correspondence field estimation and pose refinement are conducted alternatively in each iteration to recover the object poses.
Ranked #1 on 6D Pose Estimation using RGB on LineMOD
1 code implementation • ICCV 2023 • Renrui Zhang, Han Qiu, Tai Wang, Ziyu Guo, Xuanzhuo Xu, Ziteng Cui, Yu Qiao, Peng Gao, Hongsheng Li
In this paper, we introduce the first DETR framework for Monocular DEtection with a depth-guided TRansformer, named MonoDETR.
1 code implementation • 14 Mar 2022 • Fangzhou Hong, Hui Zhou, Xinge Zhu, Hongsheng Li, Ziwei Liu
In this work, we address the task of LiDAR-based panoptic segmentation, which aims to parse both objects and scenes in a unified manner.
1 code implementation • CVPR 2022 • Linjiang Huang, Liang Wang, Hongsheng Li
Our method seeks to mine the representative snippets in each video for propagating information between video snippets to generate better pseudo labels.
Pseudo Label Weakly-supervised Temporal Action Localization +1
1 code implementation • 27 Feb 2022 • Yan Xu, Junyi Lin, Jianping Shi, Guofeng Zhang, Xiaogang Wang, Hongsheng Li
The correct ego-motion estimation basically relies on the understanding of correspondences between adjacent LiDAR scans.
no code implementations • 16 Feb 2022 • Jihao Liu, Boxiao Liu, Hongsheng Li, Yu Liu
Recent studies pointed out that knowledge distillation (KD) suffers from two degradation problems, the teacher-student gap and the incompatibility with strong data augmentations, making it not applicable to training state-of-the-art models, which are trained with advanced augmentations.
Ranked #135 on Image Classification on ImageNet
no code implementations • 9 Feb 2022 • Kexue Fu, Peng Gao, Renrui Zhang, Hongsheng Li, Yu Qiao, Manning Wang
Especially, we develop a variant of ViT for 3D point cloud feature extraction, which also achieves comparable results with existing backbones when combined with our framework, and visualization of the attention maps show that our model does understand the point cloud by combining the global shape information and multiple local structural information, which is consistent with the inspiration of our representation learning method.
7 code implementations • 24 Jan 2022 • Kunchang Li, Yali Wang, Junhao Zhang, Peng Gao, Guanglu Song, Yu Liu, Hongsheng Li, Yu Qiao
Different from the typical transformer blocks, the relation aggregators in our UniFormer block are equipped with local and global token affinity respectively in shallow and deep layers, allowing to tackle both redundancy and dependency for efficient and effective representation learning.
Ranked #153 on Image Classification on ImageNet
2 code implementations • 12 Jan 2022 • Kunchang Li, Yali Wang, Peng Gao, Guanglu Song, Yu Liu, Hongsheng Li, Yu Qiao
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.
no code implementations • 11 Jan 2022 • Zipeng Qin, Jianbo Liu, Xiaolin Zhang, Maoqing Tian, Aojun Zhou, Shuai Yi, Hongsheng Li
The recently proposed MaskFormer gives a refreshed perspective on the task of semantic segmentation: it shifts from the popular pixel-level classification paradigm to a mask-level classification method.
2 code implementations • CVPR 2022 • Renrui Zhang, Ziyu Guo, Wei zhang, Kunchang Li, Xupeng Miao, Bin Cui, Yu Qiao, Peng Gao, Hongsheng Li
On top of that, we design an inter-view adapter to better extract the global feature and adaptively fuse the few-shot knowledge learned from 3D into CLIP pre-trained in 2D.
Ranked #3 on 3D Open-Vocabulary Instance Segmentation on STPLS3D
3D Open-Vocabulary Instance Segmentation Few-Shot Learning +6
1 code implementation • CVPR 2022 • Xizhou Zhu, Jinguo Zhu, Hao Li, Xiaoshi Wu, Xiaogang Wang, Hongsheng Li, Xiaohua Wang, Jifeng Dai
The model is pre-trained on several uni-modal and multi-modal tasks, and evaluated on a variety of downstream tasks, including novel tasks that did not appear in the pre-training stage.
1 code implementation • NeurIPS 2021 • Wei Sun, Aojun Zhou, Sander Stuijk, Rob Wijnhoven, Andrew Oakleigh Nelson, Hongsheng Li, Henk Corporaal
However, the existing N:M algorithms only address the challenge of how to train N:M sparse neural networks in a uniform fashion (i. e. every layer has the same N:M sparsity) and suffer from a significant accuracy drop for high sparsity (i. e. when sparsity > 80\%).
2 code implementations • NeurIPS 2021 • Peng Gao, Jiasen Lu, Hongsheng Li, Roozbeh Mottaghi, Aniruddha Kembhavi
Convolutional neural networks (CNNs) are ubiquitous in computer vision, with a myriad of effective and efficient variations.
1 code implementation • 29 Nov 2021 • Teli Ma, Shijie Geng, Mengmeng Wang, Jing Shao, Jiasen Lu, Hongsheng Li, Peng Gao, Yu Qiao
Recent advances in large-scale contrastive visual-language pretraining shed light on a new pathway for visual recognition.
Ranked #4 on Long-tail Learning on Places-LT (using extra training data)
1 code implementation • CVPR 2022 • Yi Zhang, Dasong Li, Ka Lung Law, Xiaogang Wang, Hongwei Qin, Hongsheng Li
To evaluate raw image denoising performance in real-world applications, we build a high-quality raw image dataset SenseNoise-500 that contains 500 real-life scenes.
1 code implementation • 6 Nov 2021 • Renrui Zhang, Rongyao Fang, Wei zhang, Peng Gao, Kunchang Li, Jifeng Dai, Yu Qiao, Hongsheng Li
To further enhance CLIP's few-shot capability, CLIP-Adapter proposed to fine-tune a lightweight residual feature adapter and significantly improves the performance for few-shot classification.
1 code implementation • ICCV 2021 • Yi Zhang, Hongwei Qin, Xiaogang Wang, Hongsheng Li
However, the real raw image noise is contributed by many noise sources and varies greatly among different sensors.
Ranked #2 on Image Denoising on SID SonyA7S2 x100
2 code implementations • 9 Oct 2021 • Peng Gao, Shijie Geng, Renrui Zhang, Teli Ma, Rongyao Fang, Yongfeng Zhang, Hongsheng Li, Yu Qiao
Large-scale contrastive vision-language pre-training has shown significant progress in visual representation learning.
no code implementations • 8 Oct 2021 • Jihao Liu, Hongsheng Li, Guanglu Song, Xin Huang, Yu Liu
Recently, transformer and multi-layer perceptron (MLP) architectures have achieved impressive results on various vision tasks.
Ranked #239 on Image Classification on ImageNet
3 code implementations • ICLR 2022 • Kunchang Li, Yali Wang, Gao Peng, Guanglu Song, Yu Liu, Hongsheng Li, Yu Qiao
For Something-Something V1 and V2, our UniFormer achieves new state-of-the-art performances of 60. 8% and 71. 4% top-1 accuracy respectively.
Ranked #7 on Action Recognition on Something-Something V1
1 code implementation • 12 Sep 2021 • Xinge Zhu, Hui Zhou, Tai Wang, Fangzhou Hong, Wei Li, Yuexin Ma, Hongsheng Li, Ruigang Yang, Dahua Lin
In this paper, we benchmark our model on these three tasks.
1 code implementation • ICCV 2021 • Rui Liu, Hanming Deng, Yangyi Huang, Xiaoyu Shi, Lewei Lu, Wenxiu Sun, Xiaogang Wang, Jifeng Dai, Hongsheng Li
On the contrary, the soft composition operates by stitching different patches into a whole feature map where pixels in overlapping regions are summed up.
Ranked #3 on Video Inpainting on DAVIS
1 code implementation • ICCV 2021 • Ziniu Wan, Zhengjia Li, Maoqing Tian, Jianbo Liu, Shuai Yi, Hongsheng Li
To this end, we propose Multi-level Attention Encoder-Decoder Network (MAED), including a Spatial-Temporal Encoder (STE) and a Kinematic Topology Decoder (KTD) to model multi-level attentions in a unified framework.
Ranked #39 on 3D Human Pose Estimation on MPI-INF-3DHP
no code implementations • 19 Aug 2021 • Ning Wang, Guangming Zhu, Liang Zhang, Peiyi Shen, Hongsheng Li, Cong Hua
With the effective spatio-temporal relationship modeling, it is possible not only to uncover contextual information in each frame but also to directly capture inter-time dependencies.
1 code implementation • ICCV 2021 • Xiaoyang Guo, Shaoshuai Shi, Xiaogang Wang, Hongsheng Li
Compared with the state-of-the-art stereo detector, our method has improved the 3D detection performance of cars, pedestrians, cyclists by 10. 44%, 5. 69%, 5. 97% mAP respectively on the official KITTI benchmark.
no code implementations • 17 Aug 2021 • Lin Zhao, Hui Zhou, Xinge Zhu, Xiao Song, Hongsheng Li, Wenbing Tao
However, two major issues of the fusion between camera and LiDAR hinder its performance, \ie, how to effectively fuse these two modalities and how to precisely align them (suffering from the weak spatiotemporal synchronization problem).
no code implementations • 15 Aug 2021 • Jihan Yang, Shaoshuai Shi, Zhe Wang, Hongsheng Li, Xiaojuan Qi
These specific designs enable the detector to be trained on meticulously refined pseudo labeled target data with denoised training signals, and thus effectively facilitate adapting an object detector to a target domain without requiring annotations.
1 code implementation • ICCV 2021 • Linjiang Huang, Liang Wang, Hongsheng Li
In this paper, we present a framework named FAC-Net based on the I3D backbone, on which three branches are appended, named class-wise foreground classification branch, class-agnostic attention branch and multiple instance learning branch.
1 code implementation • ICCV 2021 • Peng Gao, Minghang Zheng, Xiaogang Wang, Jifeng Dai, Hongsheng Li
However, DETR suffers from its slow convergence.
1 code implementation • ICCV 2021 • Zhipeng Luo, Zhongang Cai, Changqing Zhou, Gongjie Zhang, Haiyu Zhao, Shuai Yi, Shijian Lu, Hongsheng Li, Shanghang Zhang, Ziwei Liu
In addition, existing 3D domain adaptive detection methods often assume prior access to the target domain annotations, which is rarely feasible in the real world.
1 code implementation • 7 Jul 2021 • Xiaohan Xing, Yuenan Hou, Hang Li, Yixuan Yuan, Hongsheng Li, Max Q. -H. Meng
With the contribution of the CCD and CRP, our CRCKD algorithm can distill the relational knowledge more comprehensively.
1 code implementation • 2 Jul 2021 • Jiahui Li, Wen Chen, Xiaodi Huang, Zhiqiang Hu, Qi Duan, Hongsheng Li, Dimitris N. Metaxas, Shaoting Zhang
To handle this problem, we propose a hybrid supervision learning framework for this kind of high resolution images with sufficient image-level coarse annotations and a few pixel-level fine labels.
1 code implementation • CVPR 2021 • Xiao Zhang, Yixiao Ge, Yu Qiao, Hongsheng Li
Unsupervised object re-identification targets at learning discriminative representations for object retrieval without any annotations.
no code implementations • 4 Jun 2021 • Peng Gao, Shijie Geng, Yu Qiao, Xiaogang Wang, Jifeng Dai, Hongsheng Li
In this paper, we propose a novel Scalable Transformers, which naturally contains sub-Transformers of different scales and have shared parameters.
4 code implementations • 2 Jun 2021 • Peng Gao, Jiasen Lu, Hongsheng Li, Roozbeh Mottaghi, Aniruddha Kembhavi
Convolutional neural networks (CNNs) are ubiquitous in computer vision, with a myriad of effective and efficient variations.
Ranked #465 on Image Classification on ImageNet
no code implementations • 25 May 2021 • Jihao Liu, Ming Zhang, Yangting Sun, Boxiao Liu, Guanglu Song, Yu Liu, Hongsheng Li
Further, an architecture knowledge pool together with a block similarity function is proposed to utilize parameter knowledge and reduces the searching time by 2 times.
1 code implementation • CVPR 2021 • Zhaoyang Huang, Han Zhou, Yijin Li, Bangbang Yang, Yan Xu, Xiaowei Zhou, Hujun Bao, Guofeng Zhang, Hongsheng Li
To address this problem, we propose a novel visual localization framework that establishes 2D-to-3D correspondences between the query image and the 3D map with a series of learnable scene-specific landmarks.
no code implementations • CVPR 2021 • Jingtan Piao, Keqiang Sun, KwanYee Lin, Quan Wang, Hongsheng Li
Since the GAR learns to model the complicated real-world image, instead of relying on the simplified graphics rules, it is capable of producing realistic images, which essentially inhibits the domain-shift noise in training and optimization.
no code implementations • 27 Apr 2021 • Yixiao Ge, Xiao Zhang, Ching Lam Choi, Ka Chun Cheung, Peipei Zhao, Feng Zhu, Xiaogang Wang, Rui Zhao, Hongsheng Li
In this way, our BAKE framework achieves online knowledge ensembling across multiple samples with only a single network.
1 code implementation • 14 Apr 2021 • Rui Liu, Hanming Deng, Yangyi Huang, Xiaoyu Shi, Lewei Lu, Wenxiu Sun, Xiaogang Wang, Jifeng Dai, Hongsheng Li
Seamless combination of these two novel designs forms a better spatial-temporal attention scheme and our proposed model achieves better performance than state-of-the-art video inpainting approaches with significant boosted efficiency.
1 code implementation • CVPR 2021 • Yingjie Cai, Xuesong Chen, Chao Zhang, Kwan-Yee Lin, Xiaogang Wang, Hongsheng Li
The key insight is that we decouple the instances from a coarsely completed semantic scene instead of a raw input image to guide the reconstruction of instances and the overall scene.
Ranked #1 on 3D Semantic Scene Completion on NYUv2
no code implementations • 7 Apr 2021 • Zhaoyang Huang, Xiaokun Pan, Runsen Xu, Yan Xu, Ka Chun Cheung, Guofeng Zhang, Hongsheng Li
However, local image contents are inevitably ambiguous and error-prone during the cross-image feature matching process, which hinders downstream tasks.
1 code implementation • 5 Apr 2021 • Yunhe Gao, Rui Huang, Yiwei Yang, Jie Zhang, Kainan Shao, Changjuan Tao, YuanYuan Chen, Dimitris N. Metaxas, Hongsheng Li, Ming Chen
Radiotherapy is a treatment where radiation is used to eliminate cancer cells.
no code implementations • 31 Mar 2021 • Jiangfan Han, Mengya Gao, Yujie Wang, Quanquan Li, Hongsheng Li, Xiaogang Wang
To solve this problem, in this paper, we propose a novel student-dependent distillation method, knowledge consistent distillation, which makes teacher's knowledge more consistent with the student and provides the best suitable knowledge to different student networks for distillation.
no code implementations • CVPR 2022 • Hao Li, Tianwen Fu, Jifeng Dai, Hongsheng Li, Gao Huang, Xizhou Zhu
However, the automatic design of loss functions for generic tasks with various evaluation metrics remains under-investigated.
1 code implementation • CVPR 2021 • Rui Liu, Yixiao Ge, Ching Lam Choi, Xiaogang Wang, Hongsheng Li
Conditional generative adversarial networks (cGANs) target at synthesizing diverse images given the input conditions and latent codes, but unfortunately, they usually suffer from the issue of mode collapse.
1 code implementation • CVPR 2021 • Jihan Yang, Shaoshuai Shi, Zhe Wang, Hongsheng Li, Xiaojuan Qi
Then, the detector is iteratively improved on the target domain by alternatively conducting two steps, which are the pseudo label updating with the developed quality-aware triplet memory bank and the model training with curriculum data augmentation.
4 code implementations • ICLR 2021 • Aojun Zhou, Yukun Ma, Junnan Zhu, Jianbo Liu, Zhijie Zhang, Kun Yuan, Wenxiu Sun, Hongsheng Li
In this paper, we are the first to study training from scratch an N:M fine-grained structured sparse network, which can maintain the advantages of both unstructured fine-grained sparsity and structured coarse-grained sparsity simultaneously on specifically designed GPUs.
1 code implementation • 31 Jan 2021 • Shaoshuai Shi, Li Jiang, Jiajun Deng, Zhe Wang, Chaoxu Guo, Jianping Shi, Xiaogang Wang, Hongsheng Li
3D object detection is receiving increasing attention from both industry and academia thanks to its wide applications in various fields.
Ranked #2 on 3D Object Detection on KITTI Cars Easy val
2 code implementations • 19 Jan 2021 • Peng Gao, Minghang Zheng, Xiaogang Wang, Jifeng Dai, Hongsheng Li
The recently proposed Detection Transformer (DETR) model successfully applies Transformer to objects detection and achieves comparable performance with two-stage object detection frameworks, such as Faster-RCNN.
1 code implementation • ICCV 2021 • Chen Zhao, Yixiao Ge, Feng Zhu, Rui Zhao, Hongsheng Li, Mathieu Salzmann
Correspondence selection aims to correctly select the consistent matches (inliers) from an initial set of putative correspondences.
no code implementations • 1 Jan 2021 • Hao Shao, Yu Liu, Hongsheng Li
Inspired by spatial-based contrastive SSL, we show that significant improvement can be achieved by a proposed temporal-based contrastive learning approach, which includes three novel and efficient modules: temporal augmentations, temporal memory bank and SSTL loss.
no code implementations • 23 Dec 2020 • Daisheng Jin, Xiao Ma, Chongzhi Zhang, Yizhuo Zhou, Jiashu Tao, Mingyuan Zhang, Haiyu Zhao, Shuai Yi, Zhoujun Li, Xianglong Liu, Hongsheng Li
We observe that during training, the relationship proposal distribution is highly imbalanced: most of the negative relationship proposals are easy to identify, e. g., the inaccurate object detection, which leads to the under-fitting of low-frequency difficult proposals.
no code implementations • 18 Dec 2020 • Jianbo Liu, Sijie Ren, Yuanjie Zheng, Xiaogang Wang, Hongsheng Li
With the proposed holistically-guided decoder, we implement the EfficientFCN architecture for semantic segmentation and HGD-FPN for object detection and instance segmentation.
no code implementations • 15 Dec 2020 • Jiawei Ren, Cunjun Yu, Zhongang Cai, Mingyuan Zhang, Chongsong Chen, Haiyu Zhao, Shuai Yi, Hongsheng Li
Panoptic segmentation aims at generating pixel-wise class and instance predictions for each pixel in the input image, which is a challenging task and far more complicated than naively fusing the semantic and instance segmentation results.
Ranked #11 on Panoptic Segmentation on COCO test-dev
1 code implementation • CVPR 2021 • Fangzhou Hong, Hui Zhou, Xinge Zhu, Hongsheng Li, Ziwei Liu
2) Dynamic Shifting for complex point distributions.
Ranked #2 on Panoptic Segmentation on SemanticKITTI
2 code implementations • CVPR 2021 • Xinge Zhu, Hui Zhou, Tai Wang, Fangzhou Hong, Yuexin Ma, Wei Li, Hongsheng Li, Dahua Lin
However, we found that in the outdoor point cloud, the improvement obtained in this way is quite limited.
Ranked #2 on 3D Semantic Segmentation on ScribbleKITTI
1 code implementation • 18 Nov 2020 • Minghang Zheng, Peng Gao, Renrui Zhang, Kunchang Li, Xiaogang Wang, Hongsheng Li, Hao Dong
In this paper, a novel variant of transformer named Adaptive Clustering Transformer(ACT) has been proposed to reduce the computation cost for high-resolution input.
no code implementations • 19 Oct 2020 • Yan Xu, Zhaoyang Huang, Kwan-Yee Lin, Xinge Zhu, Jianping Shi, Hujun Bao, Guofeng Zhang, Hongsheng Li
To suit our network to self-supervised learning, we design several novel loss functions that utilize the inherent properties of LiDAR point clouds.
1 code implementation • 28 Aug 2020 • Shaoshuai Shi, Chaoxu Guo, Jihan Yang, Hongsheng Li
In this technical report, we present the top-performing LiDAR-only solutions for 3D detection, 3D tracking and domain adaptation three tracks in Waymo Open Dataset Challenges 2020.
no code implementations • ECCV 2020 • Jianbo Liu, Junjun He, Jiawei Zhang, Jimmy S. Ren, Hongsheng Li
State-of-the-art semantic segmentation algorithms are mostly based on dilated Fully Convolutional Networks (dilatedFCN), which adopt dilated convolutions in the backbone networks to extract high-resolution feature maps for achieving high-performance segmentation performance.
no code implementations • 17 Aug 2020 • Rui Huang, Yuanjie Zheng, Zhiqiang Hu, Shaoting Zhang, Hongsheng Li
In most scenarios, one might obtain annotations of a single or a few organs from one training set, and obtain annotations of the the other organs from another set of training images.
3 code implementations • 4 Aug 2020 • Hui Zhou, Xinge Zhu, Xiao Song, Yuexin Ma, Zhe Wang, Hongsheng Li, Dahua Lin
A straightforward solution to tackle the issue of 3D-to-2D projection is to keep the 3D representation and process the points in the 3D space.
Ranked #11 on LIDAR Semantic Segmentation on nuScenes
1 code implementation • ECCV 2020 • Xihui Liu, Zhe Lin, Jianming Zhang, Handong Zhao, Quan Tran, Xiaogang Wang, Hongsheng Li
We propose a novel algorithm, named Open-Edit, which is the first attempt on open-domain image manipulation with open-vocabulary instructions.
1 code implementation • NeurIPS 2020 • Jiawei Ren, Cunjun Yu, Shunan Sheng, Xiao Ma, Haiyu Zhao, Shuai Yi, Hongsheng Li
In our experiments, we demonstrate that Balanced Meta-Softmax outperforms state-of-the-art long-tailed classification solutions on both visual recognition and instance segmentation tasks.
Ranked #7 on Long-tail Learning on CIFAR-10-LT (ρ=10)
no code implementations • 20 Jul 2020 • Haisheng Su, Jinyuan Feng, Hao Shao, Zhenyu Jiang, Manyuan Zhang, Wei Wu, Yu Liu, Hongsheng Li, Junjie Yan
Specifically, in order to generate high-quality proposals, we consider several factors including the video feature encoder, the proposal generator, the proposal-proposal relations, the scale imbalance, and ensemble strategy.
2 code implementations • ECCV 2020 • Xiaokang Chen, Kwan-Yee Lin, Jingbo Wang, Wayne Wu, Chen Qian, Hongsheng Li, Gang Zeng
Depth information has proven to be a useful cue in the semantic segmentation of RGB-D images for providing a geometric counterpart to the RGB representation.
no code implementations • 8 Jul 2020 • Shijie Geng, Peng Gao, Moitreya Chatterjee, Chiori Hori, Jonathan Le Roux, Yongfeng Zhang, Hongsheng Li, Anoop Cherian
Given an input video, its associated audio, and a brief caption, the audio-visual scene aware dialog (AVSD) task requires an agent to indulge in a question-answer dialog with a human about the audio-visual content.
2 code implementations • 16 Jun 2020 • Siyu Chen, Junting Pan, Guanglu Song, Manyuan Zhang, Hao Shao, Ziyi Lin, Jing Shao, Hongsheng Li, Yu Liu
This technical report introduces our winning solution to the spatio-temporal action localization track, AVA-Kinetics Crossover, in ActivityNet Challenge 2020.
3 code implementations • CVPR 2021 • Junting Pan, Siyu Chen, Mike Zheng Shou, Yu Liu, Jing Shao, Hongsheng Li
We propose to explicitly model the Actor-Context-Actor Relation, which is the relation between two actors based on their interactions with the context.
Ranked #2 on Action Recognition on AVA v2.1
3 code implementations • ECCV 2020 • Yixiao Ge, Haibo Wang, Feng Zhu, Rui Zhao, Hongsheng Li
The task of large-scale retrieval-based image localization is to estimate the geographical location of a query image by recognizing its nearest reference images from a city-scale dataset.
3 code implementations • NeurIPS 2020 • Yixiao Ge, Feng Zhu, Dapeng Chen, Rui Zhao, Hongsheng Li
To solve these problems, we propose a novel self-paced contrastive learning framework with hybrid memory.
Ranked #3 on Unsupervised Domain Adaptation on Market to MSMT
no code implementations • CVPR 2020 • Rui Liu, Chengxi Yang, Wenxiu Sun, Xiaogang Wang, Hongsheng Li
Large-scale synthetic datasets are beneficial to stereo matching but usually introduce known domain bias.
no code implementations • ECCV 2020 • Jianbo Liu, Junjun He, Jimmy S. Ren, Yu Qiao, Hongsheng Li
Long-range contextual information is essential for achieving high-performance semantic segmentation.
2 code implementations • CVPR 2020 • Xiaokang Chen, Kwan-Yee Lin, Chen Qian, Gang Zeng, Hongsheng Li
To this end, we first propose a novel 3D sketch-aware feature embedding to explicitly encode geometric information effectively and efficiently.
3D Semantic Scene Completion from a single RGB image Hallucination
4 code implementations • 14 Mar 2020 • Yixiao Ge, Feng Zhu, Dapeng Chen, Rui Zhao, Xiaogang Wang, Hongsheng Li
To tackle the challenges, we propose an end-to-end structured domain adaptation framework with an online relation-consistency regularization term.
Ranked #4 on Unsupervised Domain Adaptation on Market to MSMT
1 code implementation • 25 Feb 2020 • Yushi Lan, Yu-An Liu, Maoqing Tian, Xinchi Zhou, Xuesen Zhang, Shuai Yi, Hongsheng Li
Meanwhile, we introduce "Semantic Fusion Branch" to filter out irrelevant noises by selectively fusing semantic region information sequentially.
no code implementations • 5 Feb 2020 • Yingjie Cai, Buyu Li, Zeyu Jiao, Hongsheng Li, Xingyu Zeng, Xiaogang Wang
Monocular 3D object detection task aims to predict the 3D bounding boxes of objects based on monocular RGB images.
1 code implementation • 30 Jan 2020 • Liang Zhang, Xudong Wang, Hongsheng Li, Guangming Zhu, Peiyi Shen, Ping Li, Xiaoyuan Lu, Syed Afaq Ali Shah, Mohammed Bennamoun
To solve these problems mentioned above, we propose a novel graph self-adaptive pooling method with the following objectives: (1) to construct a reasonable pooled graph topology, structure and feature information of the graph are considered simultaneously, which provide additional veracity and objectivity in node selection; and (2) to make the pooled nodes contain sufficiently effective graph information, node feature information is aggregated before discarding the unimportant nodes; thus, the selected nodes contain information from neighbor nodes, which can enhance the use of features of the unselected nodes.
2 code implementations • ICLR 2020 • Yixiao Ge, Dapeng Chen, Hongsheng Li
In order to mitigate the effects of noisy pseudo labels, we propose to softly refine the pseudo labels in the target domain by proposing an unsupervised framework, Mutual Mean-Teaching (MMT), to learn better features from the target domain via off-line refined hard pseudo labels and on-line refined soft pseudo labels in an alternative training manner.
12 code implementations • CVPR 2020 • Shaoshuai Shi, Chaoxu Guo, Li Jiang, Zhe Wang, Jianping Shi, Xiaogang Wang, Hongsheng Li
We present a novel and high-performance 3D object detection framework, named PointVoxel-RCNN (PV-RCNN), for accurate 3D object detection from point clouds.
no code implementations • ICCV 2019 • Yan Xu, Xinge Zhu, Jianping Shi, Guofeng Zhang, Hujun Bao, Hongsheng Li
Most of existing methods directly train a network to learn a mapping from sparse depth inputs to dense depth maps, which has difficulties in utilizing the 3D geometric constraints and handling the practical sensor noises.
1 code implementation • NeurIPS 2019 • Xihui Liu, Guojun Yin, Jing Shao, Xiaogang Wang, Hongsheng Li
Semantic image synthesis aims at generating photorealistic images from semantic layouts.
1 code implementation • ICCV 2019 • Zihao Wang, Xihui Liu, Hongsheng Li, Lu Sheng, Junjie Yan, Xiaogang Wang, Jing Shao
Text-image cross-modal retrieval is a challenging task in the field of language and vision.
Ranked #9 on Image Retrieval on Flickr30K 1K test
no code implementations • ICCV 2019 • Jiageng Mao, Xiaogang Wang, Hongsheng Li
Our InterpConv is shown to be permutation and sparsity invariant, and can directly handle irregular inputs.
Ranked #27 on 3D Part Segmentation on ShapeNet-Part
no code implementations • ICCV 2019 • Peng Gao, Haoxuan You, Zhanpeng Zhang, Xiaogang Wang, Hongsheng Li
The proposed module learns the cross-modality relationships between latent visual and language summarizations, which summarize visual regions and question into a small number of latent representations to avoid modeling uninformative individual region-word relations.
no code implementations • 28 Jul 2019 • Yunhe Gao, Rui Huang, Ming Chen, Zhe Wang, Jincheng Deng, YuanYuan Chen, Yiwei Yang, Jie Zhang, Chanjuan Tao, Hongsheng Li
In this paper, we propose an end-to-end deep neural network for solving the problem of imbalanced large and small organ segmentation in head and neck (HaN) CT images.
1 code implementation • 9 Jul 2019 • Jiahui Li, Shuang Yang, Xiaodi Huang, Qian Da, Xiaoqun Yang, Zhiqiang Hu, Qi Duan, Chaofu Wang, Hongsheng Li
Our framework achieves accurate signet ring cell detection and can be readily applied in the clinical trails.
6 code implementations • 8 Jul 2019 • Shaoshuai Shi, Zhe Wang, Jianping Shi, Xiaogang Wang, Hongsheng Li
3D object detection from LiDAR point cloud is a challenging problem in 3D scene understanding and has many practical applications.