pytorch-detect-rfcn
pytorch-detect-rfcn copied to clipboard
Feynman27
Pytorch implementation of RFCN used as baseline for Imagenet VID+DET in https://arxiv.org/abs/1710.03958.
it seems that only imagenet mean was used without std. and images were not normalized by 255. Do these steps matter in preprocessing? or make little difference.
thank you for your work! I run the train code, but the memory continues to rise. At last, it is out of memory.  How to handle this?
Hi, I met a strange problem when using mGPUs for traning, which says: RuntimeError: arguments are located on different GPUs at /opt/conda/conda-bld/pytorch_1513363039688/work/torch/lib/THC/generated/../generic/THCTensorMathPointwise.cu:269 The configs are as followed: Namespace(batch_size=4, checkepoch=1, checkpoint=0,...
Hi, sorry for bothering. I met a problem when run the code trainval_net.py. with following cfgs Namespace(batch_size=2, checkepoch=1, checkpoint=0, checkpoint_interval=10000, checksession=1, class_agnostic=False, cuda=True, dataset='imagenet_vid+imagenet_det', disp_interval=100, large_scale=False, lr=0.001, lr_decay_gamma=0.1, lr_decay_step=5, mGPUs=True,...
I [vid_eval](https://github.com/Feynman27/pytorch-detect-rfcn/blob/master/lib/datasets/vid_eval.py), can you explain what ```det = [False] * len(R)``` does exactly? and why we have it here?
