Task –
Train your own image classification model with at least 3 classes and test it in the browser.
My Lecturer
Delving into data and machine intelligence I had a go at training an image classification model. My model had four classes and its main goal was to identify if the user was making the letter A, B or C with their hands. (The letters follow the American sign language Alphabet).
This process definitely had some trial and error as the model checks for a variety of factors when assessing what qualifies for a match. When checking for matches it looks for similarities in background and colours etc. so I had to take images in different areas to add to the data pool. However, I should have added images of other people with different features as I found out when other people tested the model the results weren’t as accurate. In the future, I would like to add images of a variety of different people and locations to have a more rounded data pool.
Results of My Model ↓
Link to the model – https://teachablemachine.withgoogle.com/models/n_kDDo8SA/
