Arhat is a cross-platform, deep learning framework that converts neural network descriptions into lean, stand-alone executable code. This approach provides significant benefits because of a simple and straightforward deployment process.

Arhat is integrated with Intel® oneAPI Deep Neural Network Library (oneDNN). The Arhat back end for Intel platforms generates C++ code that directly calls oneDNN. Furthermore, Arhat provides a module that consumes models produced by the OpenVINO™ toolkit model optimizer.

This presentation shows recent case studies dedicated to using Arhat for building object-detection applications on Intel CPU and GPU hardware. These studies cover models from the Open Model Zoo as well as models from the Detectron2 library.

Alexey Gokhberg is a seasoned software engineer with more than 25 years of experience in various industrial and academic branches. His professional interests include deep learning, high-performance computing, programming-language construction, and computational geophysics.