Settings for training the object detection model on VRD dataset (VGG16 backbone).

jz462 / Large-Scale-VRD.pytorch

Implementation for the AAAI2019 paper "Large-scale Visual Relationship Understanding"

https://arxiv.org/abs/1804.10660

MIT License

144 stars 25 forks source link

Settings for training the object detection model on VRD dataset (VGG16 backbone). #22

Closed goelarushi closed 4 years ago

goelarushi commented 4 years ago

Hi all, I have been trying to train the faster-rcnn pipeline given in this repo on the VRD dataset. I made some modifications to the config files and training code to make it fit to run for VRD. Even then my model does not seem to train and converge. I am using the SGD optimizer with lr decay and vgg16 coco pretrained weights. Any help or comments will be appreciated.

Thanks for your help in advance.