15. wildcard week
During this week at FABLab Puebla, our instructors offered three different classes tailored to our interests: robotic arms, thermoforming, and machine learning. Since I had previous experience with thermoforming, I decided to explore machine learning.
Let's begin!
Teachable Machine is a user-friendly tool for creating machine learning models for websites or apps. I am excited to experiment with this and incorporate it into future designs.
I opened the Teachable Machine website and clicked on ‘Get Started’. The site offers three types of projects: ‘Image Project’ for image-based models, ‘Audio Project’ for sound-based models, and ‘Pose Project’ for pose recognition.
I chose the 'Image Project' to create a model that can identify my two dogs, Chuby and Baco. Initially, I was skeptical since they look very similar.
The project started with two predefined classes. I renamed these to ‘Chuby’ and ‘Baco’. Using the webcam, I took pictures of Chuby and Baco, which was challenging as they kept moving. To improve the model, I also uploaded additional images I had of them.
I created a new class for ‘Humans’ and added a picture of myself along with various images of other people from the internet.
After setting up the classes and adding sufficient images, I clicked on ‘training’ to begin training the model.
Once the training was complete, I tested the model using the preview feature. The model successfully identified humans, Chuby, and Baco. I am very pleased with the outcome and look forward to integrating machine learning into future projects.
Uploading the code to my website
After training and testing the model, I proceeded to export and integrate it into my FabAcademy website. I began by clicking on the ‘Export Model’ button and selecting the ‘Tensorflow.js’ option. Then, I clicked on ‘Upload my model’ to generate the necessary JavaScript code. I copied this code and pasted it into my HTML website code. To ensure a seamless fit with my website’s layout, I made minor adjustments to the design, such as modifying the button's appearance and resizing the camera display. This allowed me to successfully incorporate the trained machine learning model into my website.
