This week, I wanted to train a model that can recognize different hand gestures. At first, I tried to make a regression model that could smoothly control the growth of a plant through gestures, but it kept failing — even with the help of AI. I’m not sure if the issue was related to the type of model I used, so I eventually switched back to a classification model. 😭 I trained three classes: closed, half, and open. Then I added some particle system effects, where each gesture corresponds to a different number and color of particles. To make the interface cleaner and more visually organized, I added a status panel to display the data collection status.
Demo:
https://drive.google.com/file/d/1bLqLaq4jU-GFrBlbV1jddvy_slNnvwC7/view?usp=sharing
Failure Record:
After switching to a regression model and making a few minor adjustments (most of the main logic stayed the same), I ended up with a very high loss, and the model couldn’t recognize my gestures in real time.
