This repository contains an experimental implementation of the YOLO-v1 object detection algorithm in PyTorch. The author replaced the original darknet backbone with ResNet50 to leverage pretrained weights. The code supports training on VOC2007+2012 datasets and evaluation, achieving 66.5% mAP at 57 FPS on a GTX 1080 Ti. It includes loss computation using autograd, data preparation scripts, and inference visualization.