FastRCNN pytorch code for object detection for PASCAL VOC dataset.
Backbone = ResNet50
Region proposals generator = edgeboxes
Run: ./download.sh to download and setup pycocotools needed for evaluation.
Run: python train.py
Put your dataset inside datasets/ folder.
This uses precomputed edgeboxes ( or any other method) as region proposals. You can change few lines in the train.py to use faster RCNN's RPN generator instead of edgeboxes or different backbone.