search

KaihuaTang / Scene-Graph-Benchmark.pytorch

A new codebase for popular Scene Graph Generation methods (2020). Visualization & Scene Graph Extraction on custom images/datasets are provided. It's also a PyTorch implementation of paper “Unbiased Scene Graph Generation from Biased Training CVPR 2020”
MIT License
1.03k stars 228 forks source link

Use of pretrained models #134

Open abhidipbhattacharyya opened 3 years ago

abhidipbhattacharyya commented 3 years ago

❓where is the PRETRAINED_DETECTOR_CKPT read during testing

Hi, First of all thank you for your well documented code. It is really nice to follow. I have following questions-

  1. When running relation_test_net.py, PRETRAINED_DETECTOR_CKPT is passed as command line argument with the path to saved model.However in side relation_test_net.py model is loaded from MODEL.WEIGHT. How then the saved model from PRETRAINED_DETECTOR_CKPT is loaded?
  2. During training how are we passing the path for pretrained FRCNN? I downloaded chekcpoints. Inside the folder there is a config.yml. There I can see the path for pretrained FRCNN. But I can not find where in code this config.yml is being used.

Thank you, Abhidip

fun1024 commented 2 years ago

hello，have you solved them?