Barath19

Barath19 / Boxer3D

Public

AR 3D object detection for iPhone with LiDAR — YOLO 2D + BoxerNet 3D lifting

3dobjectdetection arkit boxer
163
18
100% credibility
Found Apr 13, 2026 at 91 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Swift
AI Summary

An augmented reality iOS app for iPhones with LiDAR that detects common objects using AI and displays their 3D oriented bounding boxes overlaid on the camera view.

How It Works

1
📱 Discover Boxer3D

You find a cool iPhone app that uses your camera and depth sensor to spot everyday objects and wrap them in glowing 3D outlines right in the real world.

2
💻 Download the app

Grab the app files from the project page on your computer to get started.

3
📁 Add vision files

Put the special object-seeing files into the app's folder so it knows what to look for.

4
🔧 Launch on iPhone

Open the app on your iPhone 12 Pro or later, grant camera access, and point it around your room.

5
🔍 Tap to scan

Press the big blue button and feel the excitement as the app thinks for a moment.

✨ See 3D outlines appear

Colorful wireframe boxes pop up around things like chairs, bottles, or people, showing their exact size in centimeters, making the world feel magical.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 91 to 163 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Boxer3D?

Boxer3D brings AR 3D object detection to iPhone LiDAR devices, spotting everyday objects with YOLO 2D detection across 80 COCO classes, then lifting them to precise oriented 3D bounding boxes via BoxerNet. Point your camera, tap detect, and see wireframe boxes overlaid in real-time AR with labels, sizes in cm, and confidence scores—solving the pain of manual 3D measurement or basic github object recognition in augmented apps. Built in Swift with ARKit and SceneKit, it runs YOLO11n and BoxerNet ONNX models accelerated on Metal.

Why is it gaining traction?

It ports Meta's cutting-edge BoxerNet research to iOS for robust github object detection yolo, blending LiDAR depth with DINOv3 features for accurate 3D lifts where pure 2D object detection models fall short on orientation and depth. Developers dig the on-device speed, adjustable confidence thresholds, and AR rendering of object detection metrics like size and yaw, outpacing clunky object detection python ports or cloud-dependent trackers. With 86 stars, it's hooking iOS folks experimenting with object detection leaderboard tech like Boxer3D.

Who should use this?

iOS AR developers building inventory scanners, furniture placement apps, or robot perception demos needing github object tracking without extra hardware. Researchers benchmarking object detection datasets on mobile LiDAR, or indie devs prototyping object detection with deep learning a review-style pipelines for real-world object actions.

Verdict

Grab it if you're on iPhone 12 Pro+ and want a quick AR 3D detection starter—setup is straightforward via Xcode, docs cover model downloads, but 1.0% credibility and low stars signal early maturity with pending optimizations. Solid for proofs-of-concept, less for production without tests.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.