---
title: "Discussion on \"CantaStorie: Offline AI Storytelling Toy on Arduino UNO Q\" | Hashnode"
description: "Discussion on \"CantaStorie: Offline AI Storytelling Toy on Arduino UNO Q\". CantaStorie Offline AI Storytelling: CantaStorie runs YOLOX-Nano and Gemma 1B on an Arduino UNO Q to spot an object and narrate a custom story, with zero…"
url: "https://hashnode.com/posts/cantastorie-offline-ai-storytelling-toy-on-arduino-uno-q/6ac36a48575ccf237f574f5f"
---

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

Discussion

# [CantaStorie: Offline AI Storytelling Toy on Arduino UNO Q](https://circuitrocks.hashnode.dev/cantastorie-offline-ai-storytelling-toy-on-arduino-uno-q?utm_source=hashnode&utm_medium=feed)

CantaStorie Offline AI Storytelling: CantaStorie runs YOLOX-Nano and Gemma 1B on an Arduino UNO Q to spot an object and narrate a custom story, with zero…

Oct 5 · 2 min read · circuitrocks.hashnode.dev  
Read the article

-   [#arduino](https://hashnode.com/tag/arduino)
-   [#arduino-uno-q](https://hashnode.com/tag/arduino-uno-q)
-   [#edge-ai](https://hashnode.com/tag/edge-ai)
-   [#computer-vision](https://hashnode.com/tag/computer-vision)
-   [#llama-cpp](https://hashnode.com/tag/llama-cpp)

## Written by

[circuitrocks](https://hashnode.com/@circuitrocks)

@circuitrocks · Manila, Philippines

  
[View profile](https://hashnode.com/@circuitrocks)

distributor of arduino, adafruit, raspbery pi

⚡ Distributor of Open-Source Hardware 📍 Since 2015 | Metro Manila, PH 🔧 Empowering makers, students &amp; innovators

## Responses 1

Markdown supported · ⌘↵ to post

[Ahmet Özel](https://hashnode.com/@ahmetozel)

[Oct 5](https://hashnode.com/posts/cantastorie-offline-ai-storytelling-toy-on-arduino-uno-q/6ac36a48575ccf237f574f5f/comment/6ac39c1c4250b914fbc4df27)

The supported-object whitelist addresses background clutter, but I would also consider how the detector chooses one object when several whitelisted toys are visible. A high confidence score on a bear behind the child does not necessarily mean that bear is the object being presented to the camera.

A simple presentation step could require the candidate to remain near the center for several frames before confirming it through the LED ring. Measuring wrong-object selections separately from unsupported-object detections would show whether this helps. It could also avoid restarting generation whenever a briefly visible second toy changes the highest-scoring detection.
