donghyeops / graph-rcnn.pytorch

Pytorch code for our ECCV 2018 paper "Graph R-CNN for Scene Graph Generation" and other papers

graph-rcnn.pytorch (forked)

What I did

• fix out of memory(OOM) error during inference
  • remove result dictionary (result_dict keeps all results of inference for each iteration)
  • update metric score for each interation
• implement scene graph metric ('sgcls', 'predcls')
  • use gt boxes for proposals
• print triples only for --visualize

Benchmarking

Object Detection

source	backbone	model	bs	lr	lr_decay	mAP@0.5	mAP@0.50:0.95
this repo	Res-101	faster r-cnn	6	5e-3	70k,90k	24.8	12.8

Scene Graph Generation (Frequency Prior Only)

source	backbone	model	bs	lr	lr_decay	sgdet@20	sgdet@50	sgdet@100
this repo	Res-101	freq	6	5e-3	70k,90k	19.4	25.0	28.5
motifnet	VGG-16	freq	-	-	-	17.7	23.5	27.6

* freq = frequency prior baseline

Scene Graph Generation (Joint training)

source	backbone	model	bs	lr	lr_decay	sgdet@20	sgdet@50	sgdet@100
this repo	Res-101	vanilla	6	5e-3	70k,90k	10.4	14.3	16.8

Scene Graph Generation (Step training)

source	backbone	model	bs	lr	lr_decay	iter	mAP@0.5	sgdet@20	sgdet@50	sgdet@100
this repo	Res-101	vanilla	8	5e-3	20k,30k	-	24.8	10.5	13.8	16.1
this repo	Res-101	imp	8	5e-3	20k,30k	-	24.2	16.7	21.7	25.2
motifnet	VGG-16	imp	-	-	-	-	-	14.6	20.7	24.5
this repo	Res-101	msdn	8	5e-3	20k,30k	-	-	-	-	-
this repo	Res-101	grcnn	8	5e-3	20k,30k	-	-	-	-	-
my_repo	Res-101	grcnn	4	5e-3	8k,12k	15k	-	12.9	17.8	20.8