Caffe

CAFFE (Convolutional Architecture for Fast Feature Embedding) is a deep learning framework, originally developed at University of California, Berkeley. It is open source, under a BSD license.[4] It is written in C++, with a Python interface.[5]