Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

📦 ParcelVision

Mobile computer vision pipeline for 3D logistics volume estimation (m³)


🔍 Project Summary

ParcelVision set out to build a mobile CV pipeline capable of estimating the real-world 3D volume of logistics items (boxes, parcels, bags) using only a standard smartphone camera.

The core finding: True metric 3D volume from a monocular camera is mathematically infeasible without a physical reference or depth hardware. ML models can predict relative shape, but not absolute metric size — a fundamental consequence of scale ambiguity in single-camera geometry.

We successfully built a robust 2D detection, segmentation, and tracking pipeline, and exhaustively tested every monocular scale-recovery approach. The engineering foundation is production-ready and fully reusable for the recommended next steps.


✅ Pipeline — What Was Built

Stage Technology Notes
Detection YOLOv8-World Open-vocabulary, covers full logistics item vocabulary
Segmentation Segment Anything Model (SAM) Per-object pixel masks; bounding box fallback included
Tracking IoU-based temporal smoothing Stable object IDs and dimension estimates across frames
Depth Estimation DepthAnything V2 / MiDaS Relative per-frame 3D point cloud generation
IMU Sensor Fusion Gyroscope + Accelerometer + Optical Flow + Kalman Filter Fully implemented; failed due to sensor drift (see below)

🚧 The Technical Roadblock

Image pixels carry no absolute scale. A pixel could represent 1 cm or 1 m — the camera equation alone cannot resolve this.

Approaches Tested & Rejected

Approach Rejection Reason
Manual calibration Poor UX; error-prone in real conditions
Reference objects (known-size marker) Requires user placement; rejected on UX grounds
Single-frame 3D reconstruction Insufficient accuracy (outside ±10–20% target)
IMU sensor fusion Unrecoverable gyroscope + accelerometer drift on consumer hardware

None of the above achieved the required ±10–20% volumetric accuracy without user intervention that would make the app impractical.


🗺️ Recommended Next Steps

The IMU path has been exhausted. Future development must use hardware-assisted scale recovery:

Path A — LiDAR (iPhone Pro / iPad Pro)

Direct metric depth output from the LiDAR sensor eliminates scale ambiguity entirely. No estimation required.

Path B — WebXR + ARCore / ARKit

Visual Inertial Odometry via ARCore (Android) or ARKit (iOS) provides accurate metric pose estimation on supported devices without additional hardware.

Both paths are architecturally compatible with the existing pipeline. The detection, segmentation, and tracking stages are fully reusable — only the scale-recovery layer needs to be swapped in.


📁 Repository

Full technical report, architecture benchmarks, and code: Internship_Parcelvision (link to internal repo)


🛠️ Tech Stack

About

AI-powered furniture volume estimation via mobile camera using WebXR and YOLO, with automated truck selection and instant moving quotes for packers and movers.

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages