Adam had often wondered about the steps to create his own object detection machine learning model to be used on a microcontroller. How hard would it be to create one? Can simpler methods like Google’s teachable machine work for object detection? Or, will less intuitive methods like Tensorflow be more fruitful? Watch to find out what methods worked to identify soda cans and bottles.
Supplemental Content:
Generally followed this guide for the COLLAB and some Anaconda attempts: https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/training.html
Also, I followed this guide for another attempt with COLLAB and Anaconda https://github.com/EdjeElectronics/TensorFlow-Lite-Object-Detection-on-Android-and-Raspberry-Pi/blob/master/Raspberry_Pi_Guide.md
Both guides had problems with package versions for me. This is the guide I followed to have the Google Model work. The general google model I used is included in this github. https://github.com/EdjeElectronics/TensorFlow-Lite-Object-Detection-on-Android-and-Raspberry-Pi
Bill of Material:
| Product Name | Manufacturer | Quantity | Buy Kit |
|---|---|---|---|
| HD PRO WEBCAM, 3MP, 720P | LOGITECH | 1 | Buy Now |
| Raspberry Pi 4 Model B 4GB | Raspberry-Pi | 1 | Buy Now |
