• 2D image Edge detection
  _{Nankai University，Tianjin,China | 2018}
  1. Release the the official pytorch implementation code of CVPR2017 edge detention model RCF on github. 200+Star
  2. VMlab edge detection demo is online, supported by my repository
• Real time stereo matching system for capsule endoscopy
  _{Nankai University，Tianjin,China | 2019}
  1. Utilizing blender to construct a 3D gastrointestinal model and render a stereo capsule endoscopy dataset.
  2. Implementation the ECCV2018 real time stereo model StereoNet(Google, no open source) with pytorch and tensorflow.
  3. Realease the pytorch version of StereoNet on github. 200+star
  4. Improve the model performance and speed(from 60ms to 15ms)
• Self-Supervised Learning for Active Stereo Systems
  _{Nankai University，Tianjin,China | 2019}
  1. Utilizing blender to construct 3D-in-door models and render a stereo structure light dataset(Kinect dot pattern).
  2. Utilizing Intel D435 to construct a real-environment Structure light and Infrared light stereo dataset.
  3. Implementation the ECCV2018 Oral ActiveStereo model ActiveStereoNet(Google, no open source) with pytorch and tensorflow.