How harieyudha/disaster-mobilenetv2 works
Disaster-MobileNetV2 is a Python image-classification project using MobileNetV2 transfer learning to classify disaster images into six categories. Its main workflow prepares train/validation/test splits, trains a model, and evaluates it; separate modules support inference, dataset inspection, and Grad-CAM explanations. The available excerpts show dataset splitting, model construction, preprocessing, and evaluation details, but do not expose enough of the training orchestrator to verify all connections among those stages.
Main entry points: dataset.py Dataset splitting · evaluate.py Model evaluation · config.py Project configuration · train.py Training workflow
Components 15
Dataset and inputs
- Project configuration
src/config.py - Dataset splitting
src/dataset.py - Processed splits
dataset/processed - Image preprocessing
src/preprocessing.py - Training augmentation
src/augmentation.py
Model lifecycle
- Training workflow
src/train.py - MobileNetV2 classifier
src/model.py - Saved models
- Model evaluation
src/evaluate.py
Inference and analysis
- Image prediction
src/predict.py - Grad-CAM explanation
src/gradcam.py - Dataset visualization
src/visualization.py - Shared utilities
src/utils.py
Other components
- Project user
- Source images
Connections 21
- Project user → Dataset splitting: prepares data
src/dataset.py - Source images → Dataset splitting: scans images
src/dataset.py - Dataset splitting → Processed splits: copies splits
src/dataset.py - Project configuration → Dataset splitting: configures paths
src/dataset.py - Project configuration → MobileNetV2 classifier: sets parameters
src/model.py - Project configuration → Image preprocessing: sets image size
src/preprocessing.py - Project user → Training workflow: trains model
- Training workflow → Saved models: saves model
src/train.py - Project user → Model evaluation: evaluates model
src/evaluate.py - Model evaluation → Processed splits: loads test split
src/evaluate.py - Project user → Image prediction: requests prediction
src/predict.py - Image prediction → Image preprocessing: imports preprocessing
src/predict.py - Image prediction → Saved models: loads model
src/predict.py - Project user → Grad-CAM explanation: requests explanation
src/gradcam.py - Grad-CAM explanation → Model evaluation: loads model
src/gradcam.py - Grad-CAM explanation → Image preprocessing: imports preprocessing
src/gradcam.py - Project user → Dataset visualization: inspects dataset
src/visualization.py - Dataset visualization → Shared utilities: uses distribution helper
src/visualization.py - Dataset splitting → Shared utilities: uses shared helpers
src/dataset.py - Model evaluation → Shared utilities: uses shared helpers
src/evaluate.py - Training workflow → Training augmentation: uses augmentation