AI Cricket Coach is a mobile application utilizes Computer Vision to analyze practice session videos captured from the umpire’s POV, extracting key features like ball speed, line and length. Using these features, a Machine Learning model suggests the most appropriate shots for the batsman to play, improving their decision-making abilities.
This project was ranked First Place for CS Capstone Awards 2025 across all specializations (Data Analytics, Cloud Computing, Game Engineering)
The ranking was determined based on student votes as well as the Capstone Defense Evaluation, of which the rubric included:
(10%) Suitability to Solve a Real-World Problem
(22.5%) Technical Relevance – Mobile App Development
(22.5%) Technical Relevance – Machine Learning and Statistical Analysis
(22.5%) Technical Relevance – Computer Vision and Cloud Computing
(22.5%) Solution Construction Process
(10%) Solution Feasibility
https://drive.google.com/file/d/16nV_F1458UqtXBYLwEqJtoV6RG1xtwq-/view?usp=sharing
Muhammad Taimoor Aleem - Project Leader & Lead Mobile Developer
Shreya Desai - Cloud Architect & Lead Backend Developer
Feras Mahmood - Machine Learning Developer & Risk Analyst
Rashesh Desai - Computer Vision Developer
Players can record and upload delivery videos directly from the app.
The system extracts ball speed, line, length, and batsman position from uploaded videos using computer vision techniques.
A custom-trained machine learning model recommends the most appropriate batting shot based on detected ball trajectory.
Players can track their shot execution quality and average ball speed across sessions using data visualizations.
Coaches can leave feedback on each of their players' deliveries.
| Loading Page | Log In Page | Sign Up Page | Edit Profile Page |
|---|---|---|---|
 |
 |
 |
 |
| Profile Page | Home Page | Sessions History | Session Details |
|---|---|---|---|
 |
 |
 |
 |
| Ideal Shot Recommendation | Analytics Page | Coach Home Page | Add Player (Coach) |
|---|---|---|---|
 |
 |
 |
 |
| Player Comparison (Coach) |
|---|
 |
| Loading Page | Sign In Page | Sign Up Page | Forgot Password Page |
|---|---|---|---|
 |
 |
 |
 |
| Profile Page | Edit Profile | Upload Video | Active Session |
|---|---|---|---|
 |
 |
 |
 |
| Session History | Session Details | Delivery Details 1 | Delivery Details 2 |
|---|---|---|---|
 |
 |
 |
 |
| Performance Analytics |
|---|
 |
Framework: Flutter (Dart)
Architecture: Clean Architecture
Design Patterns: Repository, Singleton, Factory
Libraries: video_player, chewie, bloc, dio, SharedPreferences
Language: Python
Libraries: OpenCV, scikit-learn, imbalanced-learn, joblib
Model: RandomForestClassifier with GridSearchCV + SMOTE
Functionality: Ball trajectory extraction & ideal shot prediction
Storage: Google Cloud Storage (for video storage) & Firestore (for user, session & feedback data storage)
Serverless Compute: Cloud Run (for CV & ML inference)
Notifications: Pub/Sub
API Management: API Gateway (REST APIs for mobile app)
Copyright (c) 2025 Muhammad Taimoor Aleem
All rights reserved.
This project, including its codebase, assets, and the core idea, extraction of ball trajectory details and using them to generate ideal shot recommendation using computer vision and machine learning, is the intellectual property of Muhammad Taimoor Aleem.
Unauthorized copying, modification, distribution, use, or reuse of this software or any part of its concept, whether for academic, personal, or commercial purposes, is strictly prohibited without explicit, written permission from the copyright owner.
This repository is not open source and is provided for portfolio purposes only.
It may not be reused by:
- External developers or organizations
- Students or researchers referencing this work for academic submission
- Former or current collaborators, including capstone project teammates
Capstone project teammates are permitted to apply the technical knowledge or skills gained during the development of this project (for example in Flutter, Cloud, Machine Learning, Computer Vision) for other personal, academic, or commercial ventures, provided that the idea they pursue is substantially different from the core idea stated above.
Violation of these terms may result in legal action under applicable intellectual property laws.
For licensing inquiries or permissions, please contact: taimooraleem01@gmail.com