How pithon-s/smartguard-codebase-and-results works
SmartGuard is an experimental image-classification project, not a deployed application. Its principal workflow is preparing a local binary dataset, training one of several documented model families, and evaluating trained models; saved weights and plots capture experiment outputs. The split utility is the only sampled implementation, so notebook-to-notebook calls and exact model-to-result links are not verified. The README documents these capabilities, and the graph represents them without asserting unsupported runtime connections.
Main entry points: Results Saved model weights · split_extra_dataset_preserve_test_train.py Dataset split utility · Traning Code Files Training notebooks · Evaluation Code Files Evaluation notebooks
Components 16
Dataset preparation
- Dataset split utility
Code/Dataset Related Code Files/split_extra_dataset_preserve_test_train.py - Train/test dataset
Model training
- Training notebooks
Code/Traning Code Files - Google ViT training
Code/Traning Code Files/smart-home-google-vit.ipynb - Custom CNN training
Code/Traning Code Files/TFTrainCustomModel.ipynb - EfficientNet training
Code/Traning Code Files/TFTrainEfficientNetB0.ipynb - TensorFlow training
Code/Traning Code Files/TFTrain.ipynb
Model evaluation
- Evaluation notebooks
Code/Evaluation Code Files - Model evaluation
Code/Evaluation Code Files/EvalModel.ipynb - ViT evaluation
Code/Evaluation Code Files/EvalModelGoogleVit.ipynb - MobileNet evaluation
Code/Evaluation Code Files/EvalModelMobileNetV2.ipynb - EfficientNet evaluation
Code/Evaluation Code Files/EvalModelEfficientNetB0.ipynb
Experiment outputs
- Saved model weights
Results - Metrics and plots
Results
Other components
- Experimenter
- Local images
Connections 17
- Experimenter → Dataset split utility: runs
README.md - Local images → Dataset split utility: reads images
Code/Dataset Related Code Files/split_extra_dataset_preserve_test_train.py - Dataset split utility → Train/test dataset: copies and splits
Code/Dataset Related Code Files/split_extra_dataset_preserve_test_train.py - Experimenter → Training notebooks: runs
README.md - Train/test dataset → Training notebooks: supplies data
- Training notebooks → Saved model weights: save weights
- Experimenter → Evaluation notebooks: runs
README.md - Saved model weights → Evaluation notebooks: supplies models
- Evaluation notebooks → Metrics and plots: produce metrics
- Google ViT training → Saved model weights: save weights
- Custom CNN training → Saved model weights: save weights
- EfficientNet training → Saved model weights: save weights
- TensorFlow training → Saved model weights: save weights
- Model evaluation → Metrics and plots: produce metrics
- ViT evaluation → Metrics and plots: produce metrics
- MobileNet evaluation → Metrics and plots: produce metrics
- EfficientNet evaluation → Metrics and plots: produce metrics