Overview

The Application Developer Kit (ADK) for CEVA-XM and NeuPro streamlines the software development and integration effort required for advanced vision and AI applications. It enables entire applications to be run in a more familiar CPU environment, while automatically translating and optimizing code on the more power-efficient DSP.

The ADK helps software engineers develop computer vision applications by enabling them to optimize power efficiency and memory management, handle the complexity of imaging and vision algorithms, and simplify system-handling tasks involving frames, tiles, and so on.

Benefits

The CEVA ADK eases challenges by abstracting the use of computer vision on CEVA-XM and NeuPro based platforms directly from the CPU. Programmers benefit from using the ADK to substantially simplify and shorten the overall software development cycle and provide significant performance, memory bandwidth, and power consumption savings.

SLAM-SDK for seamless CPU offload
Easy access to OpenCV and OpenVX libraries
Abstraction of multiple functions

Main Features

  • CEVA-SLAM SDK: The Simultaneous Localization and Mapping (SLAM) software development kit accelerates SLAM development by:
    • Enabling a detailed interface from a CPU to offload the heavy lifting SLAM building blocks to the CEVA-XM and NeuPro processors
    • These building blocks utilize the DSP efficiency to support both fixed point and floating point math to extend the device’s battery life.
    • The SDK building blocks include capabilities for image processing (including Feature Detection, Feature Descriptors, Feature Matching), Linear Algebra (including Matrix Manipulation, Linear Equation Solving), Fast Sparse Equation solving for Bundle Adjustment and more.
  • CEVA-CV: A standard library of OpenCV-based functions for vision processing, fully optimized for the CEVA-XM and NeuPro family. It enables the use of pre-optimized kernels for target applications, speeding up development and optimizing performance.
  • CEVA-VX: Designed to handle all system resource requirements, including data transfers, DMA transactions, and kernel execution, thus abstracting the system architecture and automating frame handling. The CEVA-VX supports kernel tunneling, linking multiple functions, minimizing memory bandwidth, and reducing system power consumption.
  • Real-Time OS (RTOS), Scheduler: A DSP task management and scheduling software module that handles task prioritization and task switching.
  • CEVA-Link Driver: Communication channels and system drivers that completely abstract the CPU-DSP interface, and enable automatic task offloading from the CPU to the DSP.