THIS PROJECT IS MADE TO WORK ON A KHADAS VIM3 AI EDGE DEVICE.
Achived a REALTIME processing at 21 frames per second.
In this project I am deploying a YOLOV4 model for License Plate Recognition of the Indian Numbers (The ones used in the Arabic Language).
NOTE: I also send a stream of the detection results UDP port, so I can catch these packets and decode it to see a LIVE STREAM of the detection results.
The Process:
- find the car.
- find the numbers (six numbers) and order them depending on the X axis positions.
- save the detection to a MongoDB.
- snap a photo of the plate.
- compress the photo.
- encode the photo as a string and save it to the cloud.
All these results are also connected to a FULTTER app for monitoring the live stream, and cheching the results.