Basketball Tracking and Dataset Creation App

Description

A web application for basketball object detection, tracking, and automated dataset creation using YOLO object detection models. Built with FastAPI (backend) and Streamlit (frontend).

Features

• Object detection: Detect and track basketball objects including ball, player, rim, shot (made), and other basketball-related elements using YOLO models.
• Automated Dataset Generation: Create datasets from images/videos, automatically annotated in YOLO format.
• Dataset Splitting: Generate finalized datasets with customizable train, validation, and test splits, provided as JSON or downloadable ZIP files.

Class Names for YOLO Detection

The detection model (detection_model.pt) supports the following classes:

["ball", "made", "person", "rim", "shoot"]

#Setup Instructions:

 git clone <repo_link>
cd basketball_tracking_dataset_app
pip install -r requirements.txt

#Run Backend (FastAPI):

uvicorn backend.main:app --reload --port 8001

#Run Frontend (Streamlit):

streamlit run app.py

#REQUIERMENTS

---

streamlit
fastapi
uvicorn
ultralytics
opencv-python
numpy
requests
pyyaml
python-multipart

#Project Structure: