NBA AR Stats Viewer

An iOS AR app that lets fans at live NBA games (or watching without a TV) point their phone at a player, identify them using computer vision, and see their live stats overlaid in augmented reality. Built as a capstone project for an AWS internal technical community focused on spatial computing.

What It Does

Point your phone at a player during a game. Tap to screenshot. AWS Rekognition identifies the player, pulls their current stats, and the app renders them as an AR overlay floating in your camera view. Designed for in-arena fans who want instant stat access without leaving the live experience.

How It Works

1. 1.Capture — User taps on a player in the camera view, triggering a screenshot
2. 2.Identify — Screenshot is sent to AWS Rekognition which analyzes the image and identifies the player
3. 3.Fetch — Once identified, the app pulls the player's live game stats from an NBA data source
4. 4.Display — Stats are rendered as an ARKit/RealityKit overlay anchored in the camera view

• •Frontend: Native iOS built with Swift, ARKit, and RealityKit — all self-taught from scratch with no prior iOS/3D background
• •Computer Vision: AWS Rekognition for player facial/jersey recognition
• •Data Pipeline: Player identification → stats API fetch → AR rendering in real-time

Key Insight

The hardest part wasn't the AR rendering — it was bridging the gap between a noisy, low-quality arena screenshot and a confident Rekognition match. Learned that production computer vision in uncontrolled environments is a fundamentally different problem than demo-quality CV with clean inputs. Self-taught the entire iOS/ARKit/RealityKit stack from zero for this project.