Real-Time Vision System on PYNQ-Z2 FPGA | Color, Object, Face & Motion Detection in One Pipeline

In this video, I run a Real-Time Vision System on the PYNQ-Z2 — an FPGA board that uses Python-based hardware integration to process live video at the edge. The system handles multiple vision algorithms simultaneously — Color Detection, Object Detection (YOLOv3-Tiny), Face/Eye/Smile Detection, Motion Detection, Edge Detection, and Shape Detection — all running on a single pipeline with a live webcam feed, entirely implemented using the PYNQ framework and AXI DMA for high-speed data transfer.

Stack:

PYNQ-Z2 (Zynq-7000 SoC)
Python + PYNQ Framework
OpenCV 4.x
NumPy
YOLOv3-Tiny
AXI DMA

What's covered:

How the PS (ARM CPU) and PL (FPGA fabric) work together for real-time vision
Setting up the PYNQ-Z2 with webcam input
AXI DMA-based data transfer between CPU and FPGA logic
Running YOLOv3-Tiny for object detection on the board
Combining multiple vision algorithms (Color, Face, Edge, Motion, Shape) in one unified pipeline

Full source code & beginner setup guide:
https://github.com/HEMANTHKUMAR-INTI/Real-time-vision-system
#FPGA #PYNQ #PYNQZ2 #EdgeAI #ObjectDetection #YOLOv3 #FaceDetection #OpenCV #EmbeddedAI #ZynqSoC #RealTimeVision #IIIT

Видео Real-Time Vision System on PYNQ-Z2 FPGA | Color, Object, Face & Motion Detection in One Pipeline канала FPGA Works IIIT Sri City