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”
Hello and thanks for the hard work on releasing the code.
I have a question about the union feature extractor under roi_relation_feature_extractors.py. It seems that you define a few new layers in lines 48-56 and pass the stacked head/tail proposal boxes (object subject pair) to these layers. Do these layers act as some form of functional layers where you do certain transformation on them? Because I don't see any trainable weights being initialized for these layers. Thanks!
Hello and thanks for the hard work on releasing the code.
I have a question about the union feature extractor under roi_relation_feature_extractors.py. It seems that you define a few new layers in lines 48-56 and pass the stacked head/tail proposal boxes (object subject pair) to these layers. Do these layers act as some form of functional layers where you do certain transformation on them? Because I don't see any trainable weights being initialized for these layers. Thanks!
Regards, Qiang Yu