Multi Class Classification Model Training

[sharp trench]

Hi, I am trying to train a multi class classification model using the notebook provided in the template here https://docs.snap.com/lens-studio/references/templates/ml/multi-class-classification#guide

In the in the Collab Notebook there is a lot of code referring to flowers because the lens is a Flower Class detector. What if we want to use our own dataset? Do I need to go into the notebook and change the references to flowers into my dataset classes? Can this notebook be used to train my dataset?

[naive pond]

Yes the notebook examples should be possible to switch out for your own dataset.
However I would suggest you to look into Roboflow and using them to create your models or atleast use them to annotate your dataset.

[sharp trench]

I already have my images separated into corresponding folders. I need image classification instead of object detection.

[lyric lodge]

You can of course use your own classifiable images instead of flowers.

The problem with the code provided by Snap: 'Image Classification and Quantization with TensorFlow/TFLite' is that the notebook no longer works with Google Colab. Because TFLite is no longer compatible with the current version of Python. You would have to downgrade Python, which no longer works with Google Colab.
You need to use Anconda (or similar) to create an environment with an older version of Python to run the code.