A working prototype of a video monitoring system for a production line, assembled in one hour.
What the system does (visible in the video at the link):
— detection of each unit of product on the conveyor (YOLO-World: product type is specified by a text description, without weeks of dataset labeling);
— object tracking (ByteTrack) and counting through a virtual crossing line — one unit is not counted twice;
— line throughput in units/minute in real-time;
— parallel detection of people in the working area;
— automatic CSV report: time / accumulated count / rate / people in the frame.
Stack: Python, YOLO-World, ByteTrack, OpenCV. Processing — locally, without cloud services.
#yolo #computervision #python #automation #opencv
What the system does (visible in the video at the link):
— detection of each unit of product on the conveyor (YOLO-World: product type is specified by a text description, without weeks of dataset labeling);
— object tracking (ByteTrack) and counting through a virtual crossing line — one unit is not counted twice;
— line throughput in units/minute in real-time;
— parallel detection of people in the working area;
— automatic CSV report: time / accumulated count / rate / people in the frame.
Stack: Python, YOLO-World, ByteTrack, OpenCV. Processing — locally, without cloud services.
#yolo #computervision #python #automation #opencv