Original video Screenshot:
Output:
Dataset creation-Extracted images using frames from video.Created dataset of 250 Images
Created Annotations of 175 Images using “LabelImg tool”, with label as “plate”
Stored the co-ordinates of bounding box in txt file for each annotation
Trained model using YOLO Algorithm
Successfully trained the model with best Accuracy of 0.84
Also trained Tiny-yolo model for better performance & optimization, writes 10 to 11 frames in a second whereas in yolo takes 3 to 4 seconds to write a frame. Tiny-yolo gave Accuracy upto 0.82
For Auto-delete feature execution the formula used: No. of days*24*60*60
If you like this repo, please don't forget to give a ⭐.