Projects

ChromaShift: AI-Powered Color Vision Accessibility Suite

View Source
Live Preview
ChromaShift: AI-Powered Color Vision Accessibility Suite screenshot 1
ChromaShift: AI-Powered Color Vision Accessibility Suite screenshot 2
ChromaShift: AI-Powered Color Vision Accessibility Suite screenshot 3

Technology Stack

React, TypeScript, FastAPI, TensorFlow.js, ONNX Runtime, OpenCV, Docker

Project Overview

ChromaShift is an AI-powered accessibility suite that makes digital media accessible to the 300+ million individuals with Color Vision Deficiency (CVD). Unlike rigid screen filters that distort aesthetics and skin tones, ChromaShift uses a hybrid client-cloud architecture featuring a Bayesian optimization wizard for personalized calibration and semantic media remapping powered by real-time segmentation and daltonization for images, video, and documents. User evaluations validated a 95% corrected task accuracy (up from 52–76% baseline), a 25.85% reduction in completion times, high usability scores (4.21/5), and low cognitive workload (NASA-TLX 5.34/20), alongside automated WCAG 2.1 auditing with JSON report exports.

Features & Highlights

Feature 1

YOLO semantic masking engine protects natural skin tones.

Feature 2

Smart charts and graphs with dual-encoding textures for WCAG 1.4.1 compliance.

Feature 3

Non-destructive PDF vectors that preserve selectable text.