About Me

I received my PhD in Computer Science from Tsinghua University in June 2021, supervised by Prof. Haizhou Ai, and my BE in Electronic Information Engineering from Huazhong University of Science and Technology in July 2016. I joined AiFi as the Tech Lead & Engineering Manager since July 2021, before that, I worked there as an intern and Tech Lead in people tracking.

Education

Tsinghua University, Beijing, China

• PhD Candidate, Dept. of Computer Science and Technology
• Research Topic: Multi­Object Tracking, 3D Human Pose Estimation, Detection and Person Re-identification

Huazhong University of Science and Technology, Wuhan, China

• Bachelor of Electronic and Information Engineering
• Rank: 1/175

Research Projects

Multi-Person 3D Pose Estimation and Tracking from Multiple View

• Associating human keypoints across multiple cameras and frames to reconstruct and track people in 3D space. The method works in real­time and has been deployed to some realworld productions.
• Accepted by CVPR 2020 (arxiv)
• Video Demo1, Video Demo2

Online / Near-Online Multi-Object Tracking

• A series of tracking­by­detection approaches that designed for tracking multiple people in a monocular, unconstrained camera. The proposed methods achieved state­-of-­the­-art performance on MOTChallenge for both accuracy and speed at that time.
• The published papers have been cited by over 500 times.
• Paper1, Paper2, Code, Video Demo

Single Camera Relative Depth Estimation with Synthetic Data

• Obtianing relative depth from single image, trained using both real-world images from NYU Depth and synthetic images from the computer game GTA-V.
• Video1, Video2, Online Demo

Object Detection in Faster R-CNN and YOLO

• Learning PyTorch along with Faster R­CNN and YOLOv2, implemented the first, or one of the firsts, open­source object detector that written in PyTorch. The most interesting part of this project is to read the source code of Torch & PyTorch and implement the RoIPooling using Cuda.
• The repositories on Github have recieved over 2000 stars in total.
• Codes of Faster R-CNN, YOLOv2

Publications

L Chen, H Ai, R Chen, Z Zhuang, S Liu “Cross-View Tracking for Multi-Human 3D Pose Estimation at over 100 FPS”, CVPR , 2020. [pdf]

L Chen, H Ai, R Chen, Z Zhuang, “Aggregate Tracklet Appearance Features for Multi-Object Tracking”, IEEE Signal Processing Letters, 2019. [pdf]

L Chen, H Ai, Z Zhuang, C Shang, “Real-time Multiple People Tracking with Deeply Learned Candidate Selection and Person Re-Identification”, IEEE International Conference on Multimedia and Expo, 2018. [pdf] [code]

L Chen, H Ai, C Shang, Z Zhuang, B Bai, “Online Multi-Object Tracking with Convolutional Neural Networks”, IEEE International Conference on Image Processing, 2017. [pdf]

R Chen, H Ai, C Shang, L Chen, Z Zhuang, “Learning Lightweight Pedestrian Detector with Hierarchical Knowledge Distillation”, IEEE International Conference on Image Processing, 2019. [pdf] [code]

Z Zhuang, H Ai, L Chen, C Shang, “Cross-Resolution Person Re-identification with Deep Antithetical Learning”, Asian Conference on Computer Vision, 2018. [pdf]

Professional Skills

Programming Language: C/C++, Python

Deep Learning Framework: PyTorch, Tensorflow, Caffe