How sifat192/intel-image-classification-pytorch works
This repository is a notebook-based PyTorch project for classifying Intel natural-scene images into six classes. Its main workflow downloads and prepares the dataset, loads batches, trains a custom CNN, and evaluates loss and accuracy. The README documents additional training and evaluation details; the notebook is present in the tree but its contents were not sampled, so internal wiring beyond that documented workflow is not independently verified.
Main entry points: Image_Classification.ipynb Training loop · Dataset Image dataset
Components 12
Dataset Preparation
- Dataset download
notebook/Image_Classification.ipynb - Image dataset
Dataset - Image transforms
notebook/Image_Classification.ipynb - DataLoaders
notebook/Image_Classification.ipynb
Model and Training
- Custom CNN
notebook/Image_Classification.ipynb - Cross-entropy loss
notebook/Image_Classification.ipynb - Adam optimizer
notebook/Image_Classification.ipynb - Training loop
notebook/Image_Classification.ipynb
Evaluation
- Model evaluation
notebook/Image_Classification.ipynb - Loss and accuracy
notebook/Image_Classification.ipynb
Other components
- Project user
- Kaggle dataset
Connections 12
- Project user → Dataset download: runs notebook
- Dataset download → Kaggle dataset: downloads dataset
- Dataset download → Image dataset: stores images
- Image dataset → Image transforms: reads images
- Image transforms → DataLoaders: transforms images
- DataLoaders → Training loop: provides batches
- Training loop → Custom CNN: trains model
- Training loop → Cross-entropy loss: computes loss
- Training loop → Adam optimizer: updates weights
- DataLoaders → Model evaluation: provides test batches
- Model evaluation → Custom CNN: evaluates model
- Model evaluation → Loss and accuracy: reports results