xhuvom / Face-Spoof-Test

Code for Intra-dataset Benchmark in CelebA-Spoof

This is the code of inference stage for Intra-dataset Benchmark.

Dependencies

• python 3
• opencv-python
• numpy
• json
• > = Pytorch 0.4.1

Usage

• Step 1. Change line 13 - line 17 of client.py.

  #================================================================================
  # Please change following path to your OWN
  # 'metas/intra_test/test_label.json' is the label file which we have already upload in [here](https://github.com/Davidzhangyuanhan/CelebA-Spoof#dataset-downloads)
  LOCAL_ROOT = './'
  LOCAL_IMAGE_LIST_PATH = 'metas/intra_test/test_label.json'
  #================================================================================

• Step 2. Run it !!

  python main.py

• Step 3. Enjoy a cup of coffee and Wait for the output as below.

      ================================================================================
      All images finished, showing verification info below:
      ================================================================================
  INFO:root:TPR@FPR=10E-3: 0.9887188604567486
  INFO:root:TPR@FPR=5E-3: 0.9729294981691959
  INFO:root:TPR@FPR=10E-4: 0.8641818528160519
  INFO:root:Done