Home
2025Mobile Developer (Solo)Archived

Object Detection App

A cross-platform mobile app running real-time object detection entirely on-device.

Source code
iOS + Android
Platform
On-device
Inference
TFLite
Model

Problem

Cloud-based inference adds latency and a hard network dependency to what should feel like an instant camera feature. The goal was real-time object detection on a phone's live camera feed with zero cloud round-trip.

Approach

  • Built the app in Flutter for a single codebase across iOS and Android.
  • Converted a trained object detection model to TensorFlow Lite for on-device inference.
  • Optimized the camera feed pipeline to keep frame processing fast enough to feel real-time on-device.

Key decisions

TensorFlow Lite over a cloud inference API

On-device inference removes network latency and lets the app work fully offline, critical for a live camera feature.

Flutter for cross-platform delivery

One codebase for iOS and Android meant the on-device ML integration only had to be solved once.

Tech stack

FlutterTensorFlow LiteDart

Outcomes

  • Zero cloud dependency
  • Fully offline object detection running entirely on-device
  • Shipped for iOS and Android from one codebase
Back to home