Labellerr/Hands-On-Learning-in-Computer-Vision

What it solves

People wanting to learn computer vision often face a gap between theory and practical application. This repo provides ready-to-run Jupyter notebooks that demonstrate how to apply state-of-the-art models like YOLO, RT-DETR, and Depth Anything to real-world problems, such as waste classification, traffic analysis, sports analytics, industrial inspection, and agriculture monitoring.

How it works

It is a collection of notebooks, each focused on a specific use case. Each notebook typically shows how to fine-tune a pretrained model (e.g., YOLO) on a custom dataset for a particular task, and includes code cells that can be run in Jupyter or Google Colab. Many also link to a YouTube walkthrough video. The notebooks are organized under a "fine-tune YOLO for various use cases" folder, and cover tasks like object detection, instance segmentation, pose estimation, OCR, and depth-based positioning.

Who it’s for

This is for developers, students, and AI practitioners who want hands-on examples of how to apply computer vision models to practical problems. It is especially useful for those who already know some basics and want to see how to fine-tune models for custom tasks without starting from scratch.

Highlights

• A growing collection of notebooks covering many domains: waste, electronics, traffic, sports, retail, agriculture, manufacturing, and more.
• Each notebook is also accompanied by a YouTube tutorial video for step-by-step guidance.
• Most notebooks can be launched directly in Google Colab, making them easy to run without local GPU setup.
• The examples focus on fine-tuning popular models like YOLO and RT-DETR for specific tasks rather than training new models from scratch.