Basketball Game AI Critic

Automated video analysis for basketball game footage using the Twelve Labs SDK. Project Description

This project leverages the Twelve Labs SDK to provide an AI-driven analysis of basketball game videos. The script automates the process of selecting a video, uploading it to Twelve Labs, processing the video, and generating a detailed analysis. The analysis includes tracking players and the ball, evaluating shots, passes, player performance, defensive plays, game flow, crowd reactions, and broadcast data for insights into strategies and outcomes. Prerequisites

Python 3.7+
Twelve Labs API key
Git
A folder named dataset containing the basketball game videos to be analyzed

Installation

Clone the Repository:

sh

git clone https://github.com/your-username/your-repository.git cd your-repository

Install Dependencies:

sh

pip install twelvelabs

Set Up Environment Variable: Add your Twelve Labs API key to your environment variables.

sh

export TL_API_KEY=your_api_key_here

Usage

Place Videos in Dataset Folder:
Ensure your videos are in the dataset folder located in the root directory of the project. Only .mp4 files are supported.

Run the Script:
Execute the main script to start the video analysis process.

sh

python BasketballGameAICritic.py

Select a Video:
The script will display a list of available videos in the dataset folder. Enter the number corresponding to the video you want to analyze.

Monitor the Process:
The script will upload the video, process it, and generate a text analysis. Progress updates will be displayed in the terminal.

Code Overview

Here's a brief overview of the main script:

python

import os import sys import time from glob import glob from datetime import datetime from twelvelabs import TwelveLabs from twelvelabs.models.task import Task from twelvelabs.exceptions import BadRequestError

Initialize TwelveLabs client with your API key

client = TwelveLabs(api_key=os.getenv('TL_API_KEY'))

Function to dynamically select a video from the 'dataset' folder

def select_video(): video_files = glob("dataset/*.mp4") if not video_files: print("No video files found in the 'dataset' folder.") return None print("Available videos:") for i, video_file in enumerate(video_files, 1): print(f"{i}. {os.path.basename(video_file)}") choice = input("Enter the number of the video you want to analyze: ") try: choice = int(choice) if 1 <= choice <= len(video_files): return video_files[choice - 1] else: print("Invalid choice. Please enter a number within the range.") return None except ValueError: print("Invalid input. Please enter a valid number.") return None

Select the video to analyze

video_path = select_video() if not video_path: exit()

Create a unique index name based on current timestamp

index_name = f"index_{datetime.now().strftime('%Y%m%d%H%M%S')}"

Create the index with the dynamically generated name

index = client.index.create( name=index_name, engines=[ { "name": "pegasus1", "options": ["visual", "conversation"], } ] ) print(f"Index '{index_name}' created successfully.")

Define a simple loading animation

def loading_animation(): animation = "|/-\" for i in range(20): sys.stdout.write("\r" + "Processing " + animation[i % len(animation)] + " ") sys.stdout.flush() time.sleep(0.1) sys.stdout.write("\r") sys.stdout.flush()

Upload the selected video

print("Uploading video...") loading_animation() task = client.task.create(index_id=index.id, file=video_path, language="en") print("\nVideo uploaded successfully.")

Monitor the video indexing process

print("Processing video...") while True: loading_animation() try: task = client.task._get(task.id) if task.status == "ready": break except Exception as e: print(f"\nAn error occurred while retrieving task status: {e}") break print("\nVideo processing completed.")

Generate text from video

retry_count = 0 max_retries = 5 while retry_count < max_retries: try: res = client.generate.text(video_id=task.video_id, prompt="Analyze a basketball game video: track players and ball, shots made/missed, passing, player performance, defense, game flow, reactions, crowd, and broadcast data for insights into strategies and outcomes.") print(f"Comprehensive Analysis: {res.data}") break except BadRequestError as e: if "A video is not ready for generation" in str(e): print("Video not ready for text generation. Retrying...") retry_count += 1 time.sleep(10) else: print("An unexpected error occurred. Exiting.") exit()

if retry_count == max_retries: print("Reached maximum retry attempts. Please try again later.")

Close the loop for video processing

print("Video analysis completed.")

Contributing

If you would like to contribute to this project, please fork the repository and submit a pull request. For any issues, feel free to open an issue on GitHub. License

This project is licensed under the MIT License. See the LICENSE file for details. Acknowledgments

Twelve Labs for providing the API and SDK for video analysis.

Feel free to customize this README file as needed for your specific project.