How srimidhuna/intel_image_classification works
This repository combines an Intel scene-classification training workflow with a Gradio image-prediction app. The documented training path runs from labeled images through preprocessing and CNN training to evaluation and visualizations; the app loads a trained model, preprocesses an uploaded image, and returns class scores. Training details come from the README, since the notebook was not sampled; the referenced model file is not present in the supplied tree.
Main entry points: app.py Prediction and scores · code.ipynb CNN training · seg_train Training images · seg_test Test images
Components 11
Training
- Training images
intel-image-classification/seg_train/seg_train - Preprocessing and augmentation
code.ipynb - CNN training
code.ipynb - Trained model
Evaluation
- Test images
intel-image-classification/seg_test/seg_test - Classification metrics
code.ipynb - Training visualizations
code.ipynb
Prediction app
- Gradio interface
app.py - Image preprocessing
app.py - Prediction and scores
app.py
Other components
- Image user
Connections 11
- Training images → Preprocessing and augmentation: preprocesses
README.md - Preprocessing and augmentation → CNN training: feeds training
README.md - CNN training → Trained model: trains
README.md - Test images → Classification metrics: evaluates
README.md - Trained model → Classification metrics: is evaluated
README.md - CNN training → Training visualizations: visualizes training
README.md - Image user → Gradio interface: uploads image
app.py - Gradio interface → Prediction and scores: invokes
app.py - Prediction and scores → Image preprocessing: preprocesses with
app.py - Prediction and scores → Trained model: predicts with
app.py - Prediction and scores → Gradio interface: returns scores
app.py