futurely

An unofficial preliminary benchmark result is presented in https://github.com/dmlc/mxnet/issues/378#issuecomment-156730363. It is very surprising that MXNet is much faster than Caffe.

The simple factory's results are not presented there. To make sure models for both libraries are exactly the same, only the GoogLeNet and VGG 16 models from the Caffe model...

You don't have to stick with the MATLAB implementation of selective search. There's a pure [python version](https://github.com/mitmul/dlibss). Usage is a super easy one-liner `rects = dlibss.selective_search(img, 50, 200, 3, 20,...

Custom datasets need to implement their own imdb classes following the PASCAL VOC examples. https://github.com/rbgirshick/fast-rcnn/blob/master/lib/datasets/imdb.py https://github.com/rbgirshick/fast-rcnn/blob/master/lib/datasets/pascal_voc.py https://github.com/rbgirshick/fast-rcnn/blob/master/lib/datasets/factory.py

https://github.com/rbgirshick/fast-rcnn/tree/master/experiments

Will Caffe2 build a [compiler](https://developers.googleblog.com/2017/03/xla-tensorflow-compiled.html) to optimize [platform specific](https://www.tensorflow.org/performance/xla/) performace automatically?

https://caffe2.ai/docs/operators-catalogue.html#mul ```python net = core.Net('net') out = net.Mul([X, Y]) ``` https://github.com/caffe2/caffe2/search?l=Python&q=mul&type=&utf8=%E2%9C%93

What’s New in cuDNN 7? Grouped Convolutions for models such as ResNeXt and Xception and CTC (Connectionist Temporal Classification) loss layer for temporal classification https://developer.nvidia.com/cudnn

https://github.com/PointCloudLibrary/pcl/blob/master/cmake/pcl_find_cuda.cmake https://github.com/Itseez/opencv/blob/master/cmake/OpenCVDetectCUDA.cmake

@bwasti How to use the [OpenGL implementation](https://github.com/caffe2/caffe2/tree/master/caffe2/contrib/opengl) to exploit the GPUs?