This is a model used to differentiate between three types of insects:

It was created as part of the end-to-end Deep Learning Project learning process.

The model is a pretrained convnext_tiny_in22k from the timm library fine-tuned on the new dataset.

This model was trained on roughly 150 pictures of each category and performed well. However, it was not vigorously tested and so may perform badly on some edge cases images and may contain bias e.g. when training, I noticed that most images of caterpillars were next to leaves and plantation, so it may have learned to associate that environment with a caterpillar.

If you notice any weird behavior, leave a comment on the Community Tab.

I scrapped the internet for pictures of the three categories to train this model. Duckduckgo was used.

To learn how the model was trained, read this notebook.