This project intends to make trained models, such as AlexNet, GoogleNet, etc., accessible within Go. It utilizes the functionality of Intel's inference engine (part of the deep learning SDK) to interact with trained and optimized neural networks.

Note: This is work in progress.

• Go 1.8+
• Ubuntu 14.04
• Intel's Deep Learning SDK Deployment Tools
• swig 3.0.6+

• get the dlinfer package:

  go get github.com/gopherds/dlinfer
  

• declare the following environmental variable:

  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/intel/deep_learning_sdk_2016.1.0.861/deployment_tools/inference_engine/bin/intel64/lib:/opt/intel/deep_learning_sdk_2016.1.0.861/deployment_tools/inference_engine/lib/intel64
  

• build/install your Go progams as you normally would with go build and go install. See here for an example.