/Implementation-CVPR2015-CNN-for-ReID

Implementation for CVPR 2015 Paper: "An Improved Deep Learning Architecture for Person Re-Identification".

Primary LanguagePythonMIT LicenseMIT

Implementation-CVPR2015-CNN-for-ReID


Notice:

Code for CUHK03 has already been updated.

Test Environment:

  • Ubuntu 18.04
  • Python 3.6.6
  • TensorFlow 1.11
  • Keras 2.1.6-tf
  • CUDA 9.0
  • cudnn 7.3.1
  • NVIDIA GTX 1080Ti

Working on:

  • Use TensorFlow Low API to update the code
  • CMC and mAP
  • TensorBoard
  • Multi-GPU
  • Estimator

Things to do:

  • Show the Test Result Form.
  • Share the trained model to reproduce the test result with Baidu Drive or Google Drive.
  • Same to the code for Market-1501.

Implementation for CVPR 2015 Paper: "An Improved Deep Learning Architecture for Person Re-Identification".

Paper link

This architechture is implemented based on Keras with Tensorflow backen using Python Programming Language.

Train the model on which dataset?

  1. CUHK-03

The model structure