“Hey AI, which avenger am I?”, Image-aware artificial intelligence model using Teachable Machine)
Hello! AI Network Community!
Today, I would like to introduce the results that our designer and marketer made without coding. You can do everything with Ainize, an artificial intelligence developer platform using blockchain.
(It’s a project for the No-code AI Challenge developed by Julia and Lianji 🔥)
1. Project title: Hey, AI which Avenger am I?
2. Use case description
“Hey AI, which Avenger am I?” Once you upload your photo, Hey AI will analyze your face and show my similar Avengers characters along with a percentage score. This is the first edition of our no-code projects “Hey AI”.
As a marketer, a designer, and a beginner in the AI field, we have struggled to fully grasp the logic behind machine learning projects. Nowadays there are many AI no/low-code tools available, but they are still difficult for us. So, the project started with the idea, “How can we do this without knowing coding and the ins and outs of machine learning?”
We made it in two days!
After collecting data, we quickly trained an AI model on Google Teachable Machine without specific AI knowledge and built a web page with Bubble. We deployed our trained model on Ainize to use it as an API service and connected it to the no-code tool, Bubble.
In total, we only took two days to finish this project, including designing, building a model, and deploying the Web. This will be an ideal use case for beginners who just want to start their first no-code AI projects. We hope it can give them some inspiration!
3. AI model description
Our Trained Model
Teachable Machine: https://teachablemachine.withgoogle.com/models/fmTqHH1jX/
API: https://ainize.ai/woomurf/findAvenger?branch=master
Platform to build a model
We built our model using Google Teachable Machine without writing any code. It makes it super easy, fast and accessible to train and deploy ML models — no expertise or coding required.
Teachable Machine is an experiment from Google to bring a no-code and low-code approach to training AI models. Anyone with a modern browser and webcam can quickly train a model with no prior knowledge or experience with AI.
Model Architecture
Model Performance
(1) Accuracy per epoch
Accuracy: 0.911, Test Accuracy: 0.86 (After 150 epochs)
(2) Accuracy per class
(3) Confusion Matrix
4. Deployment
Deployment tool: Bubble
Deployment link: http://nocode-challenge-avengers.bubbleapps.io
Deployment description: We used the no-code tool Bubble to deploy our web. But there was a small issue. Our Teachable Machine model is compatible with Bubble only as an API; thus, we first deployed the model on Ainize as an API. And then, we connected it through Bubble’s API connector plugin.
When you export your model from Google Teachable Machine as a TensorFlow js model, you will get it written in javascript.
AI network is a blockchain-based platform and aims to innovate the AI development environment. It represents a global back-end infrastructure with millions of open source projects deployed live.
If you want to know more about us,
- AI Network website: https://ainetwork.ai/
- AI Network Official Telegram Group (English): https://t.me/ainetwork_en
- Ainize: https://ainize.ai
- AI Network YouTube: https://www.youtube.com/channel/UCnyBeZ5iEdlKrAcfNbZ-wog
- AI Network Facebook: https://www.facebook.com/ainetworkofficial
- AI Network Twitter: https://twitter.com/ai__network
