Darknet

Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.

For more information see the Darknet project website.

For questions or issues please use the Google Group.

Darknet + ZYNQ

Supports low precision training and inference on CPU and FPGA. Open a terminal on a ZCU111 board and run:

make clean
make FPGA=1

./darknet anomaly train cfg/anomaly.cfg data/ad_train.bin
./darknet anomaly test cfg/anomaly.cfg data/ad_test.bin backup/anomaly.weights

The .bit and .so files can be rebuilt here

Accelerating DNN inference and training on Zynq

Other

Language:C 76.8%Language:Tcl 14.9%Language:Cuda 6.4%Language:Python 0.8%Language:C++ 0.6%Language:Makefile 0.3%Language:Shell 0.2%