How to use it
- Download the dataset of helicopters, passenger planes and rockets.
- Fine-tune the ResNet50 model in the notebook.
- Check predictions and the confusion matrix.
How it's built
Transfer learning on a pretrained ResNet50, fine-tuned on a custom three-class image dataset, evaluated with a confusion matrix.
Why it matters for businesses
The same approach classifies any custom image categories (products, defects, documents) with little data.
In real life
A logistics company sorts package photos into damage categories using the same transfer-learning recipe.
