Custom Object Detection with YOLOv7: Dataset Curation, Training, and Real-Time Deployment

This project involves training a YOLOv7 (You Only Look Once version 7) model to detect custom objects. The process included curating a custom image dataset, performing data annotation, and fine-tuning the YOLOv7 model to achieve optimal detection accuracy.

Key Features

-Custom Dataset Curation: Collected and annotated a dataset tailored for the target objects using tools like LabelImg.

-YOLOv7 Model Training: Fine-tuned YOLOv7 on the curated dataset using PyTorch and GPU acceleration for efficient training.

-Augmentation & Preprocessing: Applied image augmentation techniques (flipping, rotation, color jittering) to improve model generalization.

-Model Optimization: Tuned hyperparameters like learning rate, batch size, and anchor boxes to enhance performance.

-Real-Time Inference: Deployed the trained model for real-time object detection using OpenCV and a custom pipeline.

Tools & Technologies Used: \ YOLOv7, PyTorch, CUDA, OpenCV LabelImg for annotation Google Colab / Local GPU for training