Zhipeng Zhang

Zhipeng Zhang

Ph.D Condidate in NLPR, CASIA. Fousing on Video object tracking/detection/segmentaion, Neural Architecture Search.

Zhipeng Zhang

I earned my Ph.D. degree from NLPR, CASIA in 2022. My current research interests encompass a range of topics, including video object tracking, monocular and multi-view 3D object detection, as well as end-to-end autonomous driving.

Email: zhangzhipeng2017@ia.ac.cn

Biography: Zhipeng received Ph.D degree from Institute of automation, Chinese Academy of Sciences (CASIA) as of 2017. Before that he received the bachelor’s degree at University of Electronic Science and Technology of China (UESTC) in 2017. He worked as an intern in Microsoft Research Asia in 2019.10-2020.6. He now works as a senior researcher at DiDi Voyager & KargoBOT.

News

  • 07/2024 CyclicRefiner for Bird’s Eye View (BEV) object detection in autonomous driving is accepted by IJCV.
  • 06/2024 VLTv2 for vision-language based single object tracking is accepted by TPAMI.
  • 06/2024 LoRAT for adapting LORA to single object tracking model is accepted by ECCV2024.
  • 03/2024 A-Teacher for semi-supervised object detection is accepted by CVPR2024.
  • 09/2023 VisEvent is finally accepted by IEEE TCYB.
  • 09/2023 BEVDist for distillating vision-based BEV object detection methods is accepted as Oral in BMVC2023.
  • 03/2023 - AUNet that designed for deepfake detection is accepted by CVPR2023.
  • 09/2022 - VLT and SwinTrack are accepted by NIPS2022.
  • 04/2022 - CNNInMo/TransInMo is accepted by IJCAI2022 as Long Oral Presentation.
  • 03/2022 - CSTrack is accepted by IEEE TIP.
  • 12/2021 - OMC/CSTrackV2:One More Check: Making” Fake Background” Be Tracked Again has been accepted by AAAI2022.
  • 09/2021 - LSCGB: Robust Texture-aware Computer-generated Image Forensic: Benchmark and Algorithm has been accepted by IEEE TIP.
  • 07/2021 - AutoMatch “Learn to Match: Automatic Matching NetworkDesign for Visual Tracking” has been accepted by ICCV2021! Preparing to release.
  • 04/2021 - SiamMOT “One More Check: Making “Fake Background” Be Tracked Again” has been released! A pretty simple but strong SOTA method for multi-object tracking (MOT).
  • 03/2021 - NLP+Tracking “Towards More Flexible and Accurate Object Tracking with Natural Language: Algorithms and Benchmark” has been accepted by CVPR2021!
  • 12/2020 - NocalSiam paper for single-object tracking (SOT) has been accepted by IEEE Trans. on Image Processing(TIP)!
  • 10/2020 - CSTrack paper for multi-object tracking (MOT) has been released!
  • 10/2020 - Ocean+ code has been released.
  • 09/2020 - Ocean+ achieves the runner-up of both VOT2020 ST (shortterm) and RT (realtime) tracks. Its variants take the 2nd/3rd/5th places of VOT2020-RT.
  • 09/2020 - SiamDW-T (submitted in VOT2019) achieves the champion of VOT2020-RGBT.
  • 08/2020 - Ocean+ is released on arxiv. A simple yet strong method for VOTS task.
  • 07/2020 - Ocean is accepted by ECCV2020. Code containing Pytorch and TensorRT versions is available.
  • 08/2019 - I achieve the runner-up in VOT2019 challenges (RGBT track). Practical collaborator of RGBD track (winner) and LT track (runner-up).
  • 03/2019 - SiamDW is accepted by CVPR2019 as an Oral. Code is available.