KO
|
EN
gitlite — search
Search
#golang
#python
#java
#api
#android
#javascript
#kubernetes
#ios
#flutter
#morphology
#hacktoberfest
#tutorial
genai-tm
★ 37
Open GitHub ↗
GenAI Teachable Machine
Download README (.md)
Explore Similar Repositories
flutter_tflite_audio
:
Audio classification Tflite package for flutter (iOS & Android). Can support Google Teachable Machine models
Image-Classification-App-with-Teachable-Machine
:
Android app that uses a TensorFlow Lite model for image classification of common objects, trained through Google's Teachable Machine.
teachablemachine-node
:
Using Teachable Machine Models in Node.js
A.I.-Shooting-Game-ML-Agents-Unity-Example
:
A beginner friendly example for Unity's ML-Agents Framework. This project teaches you how to train an A.I. via Machine Learning.
project-teachable
:
Example Project: Teachable Machine
// repository documentation
Was this content helpful?
★ 0
(0 ratings)
Select Rating:
★
★
★
★
★
Submit Feedback
Recent Feedback
×
Download README
Do you want to download the
README.md
file for
genai-tm
?
Download (.md)
# GenAI Teachable Machine A simple web-based Machine Learning environment for children and learners to create their own image classifier with behaviours. This variation does not require coding skills, it provides a way to add actions and behaviours to the model without block-based or scripting languages. A version of this application is freely available [here](https://tm.gen-ai.fi/). The site does not track you or send any data to servers, all calculations are done in your browser. The sharing and collaboration functionality is peer-2-peer which requires our server to initiate the connection. ## Who made it? This application is developed as a part of the [Generation AI](https://www.generation-ai-stn.fi) research project in Finland, in partnership with 12 schools. ### Citation _N. Pope, J. Kahila, H. Vartiainen and M. Tedre, "Children's AI Design Platform for Making and Deploying ML-Driven Apps: Design, Testing, and Development," in IEEE Transactions on Learning Technologies, doi: 10.1109/TLT.2025.3529994._ ## What can it do? The classifier uses a deep neural network called MobileNet (version 2) which has been pre-trained on the ImageNet dataset. The app then uses transfer learning to fine tune the classification result to custom classes. Since the trained MobileNet model is not modified by the application, it can only work well with the types of images found in the ImageNet dataset. It cannot work well for recognising people, faces or emotions but is good at distinguishing different kinds of objects. ## Installation This is a [React](https://react.dev/) web application that is developed within Node.js using `npm`. If you wish to build your own deployment you will need Node.js installed on your machine. Steps to install and build: 1. Download the source code from Github. 2. `npm install` 3. `npm run build` 4. Copy the contents of the `dist` folder to a web server. For the Peer-2-Peer functionality, some environment variables need to be configured to point to your own server. Please contact us if you need help creating your own server, based around the PeerJS package. ``` VITE_APP_PEER_SERVER= VITE_APP_PEER_SECURE=1 VITE_APP_PEER_PORT=443 VITE_APP_PEER_KEY= VITE_APP_APIURL= ``` ## Development The app uses "vite". You can start a development server using `npm start` and use `npm test` to run the automated tests.
