resnet50onDogsVsCatsDataset

This example shows how to do image classification from scratch using pre-made resnet50Light model, without leveraging pre-trained weights. We demonstrate the workflow on the Kaggle Cats vs Dogs binary classification dataset.

We use the Preprocessing DSL to describe the dataset generation pipeline.

It includes:

• dataset loading from S3

• preprocessing DSL declaration

• OnFlyImageDataset dataset creation

• dataset splitting

• usage of pre-made model from org.jetbrains.kotlinx.dl.api.core.model package

• model compilation

• model training

• model evaluation