Barcode detection and reading is an application typically ran using an accelerated industrial PC running a mix of traditional and AI/ML based algorithms. As designers look to reduce the cost of deployment, cameras with built-in barcode reading capabilities enable simpler implementation of the application.
This reference design demonstrates the barcode detection using the CertusPro-NX Voice and Vision Machine Learning Board that implements camera barcode detection capabilities based on Yolov5 NN models.
Features
- Detection, image quality improvement and reading in a single design
- Low latency design based on the parallel compute capabilities of the CertusPro-NX FPGA
- Image resolution input of 640 x 480, cropped and downscaled to 160 x 120 for processing
- Detection results overlayed within the FPGA on top of the input image
- Uses Yolov5 style network for improved accuracy in small footprint
