search

jwyang / graph-rcnn.pytorch

[ECCV 2018] Official code for "Graph R-CNN for Scene Graph Generation"
732 stars 157 forks source link

VGG Checkpoint #25

Open alirezazareian opened 5 years ago

alirezazareian commented 5 years ago

Thanks for this great repository.

In the ECCV paper, you used a VGG backbone. Could you please share its checkpoints? I understand you have provided a better ResNet backbone already, but I need to compare with the numbers you reported in the paper, and hence I need the original model's checkpoint.

Thanks very much for helping.

jwyang commented 5 years ago

Yes, I will also train with VGG and other backbones, will let you know when it finished.

jnhwkim commented 5 years ago

I believe @Jaesuny is working now to exploit MotifNet's pre-trained parameters. A couple of days is needed. Stay tuned!

Scarecrow0 commented 4 years ago

Hi, Is VGG16 backbone ready? Seems that the VGG backbone are not in the dev plan of maskrcnn-benchmark, but VGG is the most important backbone for Scene graph generation. It will be better if we have a classic VGG backbone in this great repo which has various of SGG pipeline.

Thanks !

emasa commented 4 years ago

This is a great repository! Thanks for sharing!

For a fair comparison, I need to compare with the results reported in the paper. Would it be possible to get access to the VGG16 checkpoint? If that's too inconvenient, it'd be really helpful to have the VGG16 config files to train the model from scratch! Thanks for your help!