Raghavendra Kotikalapudi

Sure. Where should the tests go? would resnet.py go into applications?

Looks very good. I honestly don't have time to maintain this repo anymore. I will simply redirect users to your repo for latest changes.

How about trying dilations instead? It precisely solves this very issue: https://arxiv.org/pdf/1511.07122.pdf

True..I think i should just leave out the dense layer as flexible since it can be used for regression as well.

cool

I don't think all the subtleties in section 4.2 are implemented (https://arxiv.org/pdf/1512.03385.pdf) This is just a sample application. Feel free to submit a PR. It would be good to have...

For the first one, it looks like it might be overfitting after epoch 30 ish... I would try lowering learning rate. The second one looks better imo. Did you drop...

Doesn't that require pre-trained weights? I can work on porting over resnet50 weights. Not sure about others. Do you know if Kaiming He has pretrained weights for all resnet configurations?...

Ok. I'll give it a shot. Do you know of any projects for converting caffe models? I see a bunch of them floating around. Not sure which one to use.

Cool. I will try to get to this in after december. In the mean time, if any of you guys have time, feel free to take a stab at this....