Object Classification Demonstration

CNN accelerated object classification using Low Power Lattice FPGAs

Object classification and tracking applications are used across markets. This demonstration provides a sample application for detecting, classifying, and tracking multiple objects running on a low power, general purpose Lattice CertusPro-NX FPGA.

Demo design is provided with end-to-end components to allow for modification of the object being detected.

Video Streaming Analysis in Avant-E platform – Able to see detection boxes. The Green boxes detect vehicles whereas Blue detect Person.

CertusPro-NX MobileNet Object Classification – Implements generic object classification base design using the CertusPro™-NX platform.

Features

  • Accelerated, low-power object classification demonstration
  • Wide range of applications including defect detection and robot navigation
  • Processing at up to 33 FPS and 224x224 resolution
  • Based on Mobilenet v1 network
  • Total power consumption of CertusPro-NX FPGA is less than 400 mW

Jump to

Block Diagram

CertusPro-NX Voice and Vision ML Board Block Diagram

Documentation

*By clicking on the "Notify Me of Changes" button, you agree to receive notifications on changes to the document(s) you selected.