✈️ Airplane Detection and Tracking with YOLOv8

This project is a computer vision notebook developed and executed in a Kaggle environment, focused on automatic airplane detection and tracking in video.

The core of the project is implemented as a Jupyter Notebook, which allows the full pipeline to be run step by step: from dependency installation to video processing and final visualization of results. Kaggle is used as the execution platform due to its ease of use, reproducibility, and support for GPU-accelerated workloads.

Project Overview

The notebook implements a complete video analysis pipeline using YOLOv8, a modern real-time object detection model. The system takes a video file as input, processes it frame by frame, detects airplanes, tracks them across time, and generates a final annotated video showing bounding boxes and persistent tracking IDs.

The workflow is fully automated and includes:

• Video input loading
• Object detection focused on airplanes
• Multi-object tracking to maintain identity across frames
• Video rendering with visual annotations
• Conversion to a final playable video format

To apply the system to a different video, the user only needs to update the video path inside the notebook. The rest of the pipeline remains unchanged.

Technologies and Tools

• Python for implementation
• YOLOv8 (Ultralytics) for object detection and tracking
• OpenCV for video handling and frame processing
• FFmpeg for video conversion
• Jupyter Notebook running on Kaggle

Example Usage

The example included in the notebook uses video footage of an F-16 aircraft as a demonstration sample. The system treats the aircraft as an airplane object and applies detection and tracking accordingly, showcasing the full pipeline behavior in a real-world scenario.

Purpose

This project serves as a practical demonstration of how modern object detection models can be integrated into a complete video processing workflow. It provides a solid foundation for experimenting with computer vision, object tracking, and real-time video analysis in a reproducible notebook-based environment.