---
title: "Discussion on \"I Turned 149k Messy Images into a Recognition System That Works Offline\" | Hashnode"
description: "Discussion on \"I Turned 149k Messy Images into a Recognition System That Works Offline\". Training an on-device YOLO26n food detection model from scratch across multi-source data"
url: "https://hashnode.com/posts/i-turned-149k-messy-images-into-a-recognition-system-that-works-offline/6ac741f76bd8062b27e90a2e"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [I Turned 149k Messy Images into a Recognition System That Works Offline](https://michelle-buchi.hashnode.dev/i-turned-149k-messy-images-into-a-recognition-system-that-works-offline?utm_source=hashnode&utm_medium=feed)

Training an on-device YOLO26n food detection model from scratch across multi-source data

Oct 8 · 18 min read · michelle-buchi.hashnode.dev  
Read the article

## Written by

[Michelle Buchi Okonicha](https://hashnode.com/@Michelle-Buchi)

@Michelle-Buchi

  
[View profile](https://hashnode.com/@Michelle-Buchi)

Software engineering, machine learning, web development

Software Engineer | Google Developer Expert | Women Techmakers Ambassador

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
