Food-101 PyTorch vision pipeline — 7 Steps

Recognize 10 culinary dish categories with a PyTorch computer vision pipeline: random resized crops, color jitter augmentation, MobileNetV2 backbone, and CosineAnnealingLR scheduling.

1. 01 Data ingestion & EDA
2. 02 Preprocessing without leakage
3. 03 TensorDataset & DataLoader
4. 04 nn.Module architecture
5. 05 Loss function & optimizer
6. 06 Training loop & backprop
7. 07 Eval, metrics & checkpoint