Plate number detection and money detection modesl

Good day everyone,

I want to do a project to detect vehicle license plate numbers. Do you think using YOLOv8 would be a good approach? If yes, how (should I do it)? If no, could you give me another suggestion?

Also, I’m working on a money detection model to recognize all Naira notes. However, someone might give the model a note that isn’t Naira or even money at all. Which approach should be used to detect non-Naira or non-money notes?

@paulinpaloalto @Deepti_Prasad @ai_curious @Honza_Zbirovsky @TMosh @mrtckaya please can you help in any way?

License plate reading using object detection is not a new idea. See above for one of many links easily discoverable on the interweb. Let us know how it goes.

PS: the author of that article suggest that ANPR needs to be done fast, which might be one argument for using a YOLO architecture. I suggest that is not correct; in most applications I think you only need to decide rapidly on whether or not to take a photo, which is likely done using lidar. Once you have that, you have lots of time to do the object detection and OCR. It probably doesn’t have to be done at the edge, and minutes, hours, even days to complete the downstream processing is fine.