Georgia Gkioxari

Your problems do not seem to be related to Caffe compiling issues. If compatibility was an issue you wouldn't be able to compile Caffe. I would check your hardware and...

You should test snapshots on the validation set with the test code. It is fast to test the validation set. I didn't use the validation set at all, since I...

Hi Francis, The gradient of the sum operations are straight forward. If the forward pass is f(a,b) = a+b then df/da = 1 and df/db = 1. I don't know...

You should be able to build it with the right changes in your makefile. Please refer to caffe for that...

In `src/caffe/layers/sigmoid_cross_entropy_weighted_loss_layer.cpp` try removing lines: `#ifdef CPU_ONLY STUB_GPU_BACKWARD(SigmoidCrossEntropyWeightedLossLayer, Backward); #endif`

Those fields are identical to the ones introduced by fast RCNN. Please read [the paper](http://arxiv.org/abs/1504.08083) for more information!

Check if your `fast_rcnn` directory has `config.py`. It should. If you have it, then the only reason I can think is that you don't have the right paths set up....

a. To download the Stanford 40 data, please refer to the README. It provides a link to the data b. In `lib/fast_rcnn/test_stanford40.py` in line 279 `im_detect` returns the overall score,...

This is not related to Caffe, but to Python. In `tools/train.py` in line 15 is where the dataset is imported and then called again in line 78. All you need...

For the tasks tackled in our paper, the primary region always refers to the person that is being classified. Usually the boxes of the people are provided (in order to...