This repository offers a collection of custom functions for YOLOv4 and YOLOv3 object detection models, implemented across TensorFlow, TFLite, and TensorRT runtimes. It includes utilities like object counting, detection cropping, license plate recognition using Tesseract OCR, and text extraction from detections. The project serves as an exploration of customizations and applications built on top of the YOLO architecture rather than an optimized production implementation.