Skip to content

CopymeProEIP/CopyMe

Repository files navigation

Basketball Free Throw Phases Detection

Overview

This project focuses on detecting the different phases of a basketball shot using computer vision techniques.

Download Dataset   Try Model

Tools and Technologies

  • Model: YOLOv8 by Ultralytics (link)
  • Annotation Tool: Roboflow (link)

Resources


Phases of a Basketball Shot

  1. Preparation: The player prepares to take the shot.

    preparation phase

    • Focus on the target.
    • Grip the ball properly.
    • Align the body.
  2. Position: The player gets into the correct stance and position.

    position phase

    • Feet should be shoulder-width apart.
    • Knees slightly bent.
    • Body balanced.
  3. Release: The ball is released from the player's hands.

    release phase

    • Use a smooth motion.
    • Extend the shooting arm.
    • Snap the wrist.
  4. Follow-through: The player's hands follow through after releasing the ball.

    follow through

    Be careful not to twist your wrist as you snap it forward, or else the ball may angle away from the hoop.

    • Maintain balance.
    • Keep eyes on the target.
    • Hold the follow-through position.

Steps to Train the Model

  1. Data Collection: Collect videos of basketball shots.
  2. Annotation: Use Roboflow to annotate the different phases in the videos.
  3. Training: Train the YOLOv8 model with the annotated data.
  4. Evaluation: Evaluate the model's performance on a test dataset.
  5. Deployment: Deploy the model for real-time phase detection.

Start programming in development environment with docker compose

chmod +x development.sh
./development.sh

References