How oucailab/hsi-x-classify-backbone works
HSI-X-Classify-Backbone is a multimodal remote-sensing classification experiment framework. Its principal workflow configures a dataset and model, loads and preprocesses HSI plus auxiliary-modality data, trains a registered backbone, evaluates it, and saves metrics and visualizations. The README documents the end-to-end flow and supported model and analysis capabilities; unsampled implementation details are represented only where documented, so some links cite the README.
Main entry points: model_registry.py Model registry · runner.py Experiment runner · evaluator.py Evaluator · trainer.py Training loop
Components 24
Experiment orchestration
- Experiment runner
src/runner.py - Experiment config
src/config.py - Training loop
src/trainer.py - Email notifier
src/email_notifier.py
Data preparation
- Multimodal data
src/data.py
Model backbones
- Model registry
src/model_registry.py - FusAtNet
src/models/fusatnet/model.py - ExViT
src/models/exvit/model.py - HybridSN
src/models/hybridsn/model.py - MICF-Net
src/models/micf_net/model.py - CNN
src/models/cnn/model.py - FDNet
src/models/fdnet/model.py - S2ENet
src/models/s2enet/model.py - AsyFFNet
src/models/asyffnet/model.py - HCTNet
src/models/hctnet/model.py - MACN
src/models/macn/model.py - RSCNet
src/models/rscnet/model.py
Evaluation and outputs
- Evaluator
src/evaluator.py - Classification metrics
src/metrics.py - Classification maps
src/analysis.py - t-SNE analysis
src/tsne.py - ERF visualization
src/visualize_erf.py - Experiment artifacts
Other components
- Researcher
Connections 27
- Researcher → Experiment runner: starts experiment
README.md - Experiment runner → Experiment config: uses settings
src/runner.py - Experiment runner → Training loop: runs training
src/runner.py - Experiment runner → Evaluator: runs testing
src/runner.py - Training loop → Multimodal data: loads batches
src/trainer.py - Training loop → Model registry: gets adapter
src/trainer.py - Model registry → FusAtNet: selects model
README.md - Model registry → ExViT: selects model
README.md - Model registry → HybridSN: selects model
README.md - Model registry → MICF-Net: selects model
README.md - Model registry → CNN: selects model
README.md - Model registry → FDNet: selects model
README.md - Model registry → S2ENet: selects model
README.md - Model registry → AsyFFNet: selects model
README.md - Model registry → HCTNet: selects model
README.md - Model registry → MACN: selects model
README.md - Model registry → RSCNet: selects model
README.md - Evaluator → Multimodal data: loads test data
src/evaluator.py - Evaluator → Model registry: gets adapter
src/evaluator.py - Evaluator → Classification metrics: computes metrics
src/evaluator.py - Evaluator → Classification maps: creates maps
src/evaluator.py - Evaluator → t-SNE analysis: runs analysis
src/evaluator.py - Experiment runner → Email notifier: sends notification
src/runner.py - Experiment runner → ERF visualization: enables ERF
- Classification metrics → Experiment artifacts: writes reports
README.md - Classification maps → Experiment artifacts: saves maps
README.md - t-SNE analysis → Experiment artifacts: saves plots
README.md