A Comprehensive AI Software Development Kit for Ceva-NeuPro and Ceva-SensPro processors

The Ceva-NeuPro Studio is a comprehensive AI Software Development Kit (SDK) for creating fully-optimized runtime software for Ceva-NeuPro-Nano Embedded AI NPU, Ceva-NeuPro-M NPU IP, and Ceva-SensPro sensor hub DSPs. Targeted for mass-market embedded devices, Ceva-NeuPro Studio supports multiple open AI inference frameworks, and incorporates a broad range of network optimizations, advanced quantization algorithms, data flow management and fully-optimized compute operators and libraries into a holistic solution that enables cloud-trained AI models to be deployed on resource-constrained edge devices for inference processing.

The Ceva-NeuPro Studio SDK enable an extremely simple and streamlined transition of deep neural networks to an embedded environment.
Ceva-NeuPro family NPUs ensure superior performance with minimal power consumption. Separately, each component of the Ceva-NeuPro Studio AI SDK is a powerful enabler of embedded voice, vision, imaging and sensing applications. Combined, these components deliver an ultimate toolkit to support new network structures and changing layer types of deep neural networks.

Ceva supplies a full development platform for partners and developers based on the Ceva-NeuPro-Nano, Ceva-NeuPro-M, and Ceva-SensPro architectures to enable the development of deep learning applications using the Ceva-NeuPro Studio, targeting any advanced network.

 

Benefits

The Ceva-NeuPro Studio SDK streamlines implementations of deep learning models in embedded systems by automatically quantizing and optimizing offline pre-trained neural networks to real-time embedded-ready networks for Ceva-NeuPro NPUs and Ceva-SensPro DSPs. This enables real-time, high-quality voice analytics, image classification, object recognition, and segmentation, significantly reducing time-to-market for running low power machine learning on edge devices

Automatic quantization and conversion to embedded-ready networks
Greatly reduces memory bandwidth for any network via various mechanisms including layer fusion and compression
Supports common open AI frameworks such as Caffe, TensorFlow, TensorFlow Lite Micro, TVM, and µTVM

Main Features

  • Ceva-NeuPro Studio AI SDK converts pre-trained neural network models and weights from offline training frameworks (such as Caffe or TensorFlow) to real-time network models
  • Ceva-NeuPro Studio Run-Time software accelerates deployment of machine learning in low-power embedded processors
  • Ceva-NeuPro Studio Invite API enables seamless incorporation and usage of custom AI engines

Block Diagram

​Ceva-NeuPro Studio AI SDK

 

 

Ceva-NeuPro Studio Graph Complier Demo

The Ceva-NeuPro Studio is a comprehensive graph compiler that simplifies the development and deployment of deep learning systems for mass-market embedded devices. Ceva-NeuPro Studio incorporates a broad range of network optimizations, advanced quantization algorithms, data flow management and fully-optimized CNN, RNN, and other type networks into a holistic solution that enables cloud-trained AI models to be deployed on edge devices for inference processing, using low compute and memory resources. CDNN enables heterogeneous computing and is flexible to split a network between multiple compute engines such as Ceva-SensPro or Ceva-NeuPro processors and custom AI Engines, to ensure superior performance with minimal power consumption. In this video, we are showing how the CDNN Graph Compiler and GUI enable users to quickly configure the CDNN tool and easily analyze their neural networks performance on any of Ceva’s AI processors. The example we are using is inferencing of the SSD mobilenet network on the SP500 DSP, both natively (DSP only), and also with a hardware accelerator connected via Ceva-NeuPro Studio-Invite API for higher performance.