Deeplant — Meat Data Collection
A field app for capturing labeled meat-image datasets on rugged PDA hardware.
Mobile Application Developer · Deeplant Inc. (Industry-Academic Project)
The problem
Building meat-quality models needs large, consistently labeled image datasets — but collection happened in the field on PDA devices with cameras and barcode scanners, with no reliable pipeline to get labeled images into a central store.
Goal & constraints
Deliver a Flutter app that lets field workers capture meat images tied to barcode-scanned identifiers and reliably ship them to a central server.
Key decisions
Target the PDA hardware with a single Flutter codebase.
One Flutter app could drive the PDA's camera and barcode scanner while staying maintainable for a small student-industry team.
Bridge Flutter, Flask, Firebase, and a central server over REST.
Splitting responsibilities across clear REST boundaries kept the moving parts independently debuggable across the collaboration.
How I built it
Capture app
Developed a Flutter application for collecting meat image datasets using PDA cameras and barcode scanners.
Data pipeline
Implemented REST API communication between Flutter, Flask, Firebase, and a central server to move labeled images reliably.
Team delivery
Coordinated development tasks and schedules across the engineering team to hit project milestones.
Outcome
A working field-collection pipeline that turned PDA captures into a centrally stored, barcode-labeled meat-image dataset.
What I took away
- Clear service boundaries mattered more than clever code when four systems from different teams had to interoperate.