Cricket Vision — AI Sports Analytics
Back to Projects

Cricket Vision — AI Sports Analytics

Democratizing professional-grade biomechanical analysis with WebXR

Overview

Cricket Vision brings elite-level sports analytics to the browser. Using computer vision and WebXR, it provides real-time biomechanical feedback to players and coaches, offering features previously available only in million-dollar labs.

The Challenge

Professional cricket analysis relies on expensive hardware (Hawk-Eye) and dedicated labs. Grassroots players and coaches lack access to data-driven insights about technique and mechanics.

The Solution

A browser-based AI pipeline. We use lightweight pose-estimation models (MoveNet/BlazePose) running client-side to analyze video footage. The results are overlaid in AR/WebXR, showing trajectories, angles, and speed in real-time.

Key Features

Biomechanical Pose Estimation

Tracks 33 key body points to analyze bowling action and batting stance.

High-Contrast AR Overlays

Designed specifically for outdoor sunny conditions where screens are hard to read.

Ball Trajectory Prediction

Physics engine that predicts ball path based on release velocity and spin.

Impact & Outcomes

  • Enabled remote coaching capabilities for academies.
  • Processed 10,000+ delivery videos during beta testing.
  • Featured as a top innovation in sports tech hackathon.

Tech Stack

ReactThree.jsPythonOpenCVTensorFlow.jsPostgreSQL

Implementation Hurdles

  • Running computer vision models smoothly on mobile browsers.
  • Compensating for camera shake and variable lighting in outdoor field conditions.
  • Mapping 2D video coordinates to 3D virtual space for trajectory analysis.