YOLOv1 Implementation in PyTorch

This project is a basic implementation of the YOLOv1 (You Only Look Once) object detection algorithm using PyTorch. It is designed to train on the Pascal VOC dataset and perform inference on images.

Features

• YOLOv1 Model: A convolutional neural network based on the YOLOv1 architecture (src/model.py).
• Custom Loss Function: Implementation of the YOLOv1 loss, including coordinate, objectness, no-objectness, and classification losses (src/loss.py).
• Pascal VOC Dataset Loader: Data loader specifically for the Pascal VOC dataset format (src/dataset.py).
• Training Script: Script to train the model on the VOC dataset (train.py).
• Inference Script: Script to run object detection on new images using a trained model (infer.py).
• Utilities: Includes Non-Max Suppression (NMS) (src/utils.py).

Requirements

• Python 3.x
• PyTorch
• Other dependencies listed in requirements.txt

To install the required packages, run:

pip install -r requirements.txt