Possible to load custom layers?

[danfeiX]

Would it be possible to import custom layers such as the ones in faster-rcnn? https://github.com/rbgirshick/py-faster-rcnn

Also, what if the network has two input blobs?

[szagoruyko]

this package is not for importing, see https://github.com/szagoruyko/loadcaffe FasterRCNN is not supported for loading, if you need FastRCNN then there are converted models here https://github.com/szagoruyko/fastrcnn-models.torch