nvsriram/keepervision

ECE Final Year Design Project: Using computer vision, image processing, and machine learning, KeeperVision physically guides goalkeepers in real-time to the best position to defend against any incoming shots.

💡Motivation

Soccer is the most popular sport with hundreds of millions of players and fans around the world. Many times, a key save can be the difference between winning and losing a game. KeeperVision is a training tool which aims to help goalkeepers in tackling their greatest hurdle - positioning. Using computer vision, image processing, and machine learning, KeeperVision physically guides goalkeepers in real-time to the best position to defend against any incoming shots.

🚀 Demo

KeeperVision-demo.mp4

🌐 API Reference

api/register/<username>

GET

Response	OK	Code	Content
	✅	200	{"id": <player_id>}
	❌	404	{"message": "Player <username> does not exist."}

POST

Body	{"email": <email>}
Response	OK	Code	Content
	✅	200	{"id": <player_id>}
	❌	400	{"message": "<some error message like duplicate key Integrity error>"}

api/session/<username>

GET

Response	OK	Code	Content
	✅	200	{"player_id": <player_id>, "session_stats": <list of session_stats in desc order of session_end>}
	❌	404	{"message": "Player <username> does not exist."}

POST

Body	{"session_stats": <session_stats JSON with all the fields>}
Files	File Name	Description
	initial_image	Image file containing goalkeeper's initial position before session starts
	final_image	Image file containing goalkeeper's final position at the end of session
Response	OK	Code	Content
	✅	200	{"id": <session_id>}
	❌	400	{"message": "<some error message like duplicate key Integrity error>"}
	❌	404	{"message": "Player <username> does not exist."}

api/predict

POST

Files	File Name	Description
	image	Image file to be processed
Response	OK	Code	Content
	✅	200	{"idx": <idx corresponding to direction to move>, "x": <offset in x direction>, "y": <offset in y direction>}

📈 Block Diagram

💻 Instructions on running locally

1. Ensure python is installed
2. Run pip install -r requirements.txt to install all dependencies
   NOTE: Recommended to use venv before installing dependencies. See below for details on how to set it up
3. Run the Flask app using python app.py. This will run the app on debug mode on the server's IP
   NOTE: Ensure the Flask app and the client are in the same WLAN to be able to connect to the server's IP as it is otherwise considered a private IP address

[Optional] To use venv:

1. Run python -m venv <path-to-env> to create virtual environment

2. To activate the virtual environment:

   • If on Windows, run: <path-to-env>/Scripts/activate
   • If on MacOS, run: source <path-to-env>/bin/activate

3. Run Step 2 to install all dependencies