This repository provides a clean implementation of the YOLOv3 object detection model in TensorFlow 2.0. It supports both yolov3 and yolov3-tiny variants with pre-trained Darknet weight conversion, eager and graph mode training using tf.GradientTape and model.fit, transfer learning workflows, and TensorFlow Serving deployment. The implementation follows best practices using the tf.keras functional API, tf.data pipelines, and GPU acceleration.