# deepti0611/spacenet-off-nadir architecture

> Architecture diagram of the GitHub repository deepti0611/spacenet-off-nadir, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-21). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/deepti0611/spacenet-off-nadir
- Repository: https://github.com/deepti0611/spacenet-off-nadir

## Overview

This repository is a research workflow for semantic segmentation of satellite imagery, particularly building masks. Its main stages are dataset construction and splitting, raster/mask loading and augmentation, model training, test evaluation, and result analysis. The tree also shows multiple model and training variants. Source excerpts establish a few training/evaluation dependencies; the README is empty, and many preparation and variant files were not sampled, so their connections are not asserted.

Main entry points: [`ml_pipeline/pytorch_dataset.py`](https://github.com/deepti0611/spacenet-off-nadir/blob/main/ml_pipeline/pytorch_dataset.py) (Torch dataset), [`models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py`](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py) (DeepLab training), [`models/unet_resnet34/train_unet_resnet34.py`](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/unet_resnet34/train_unet_resnet34.py) (U-Net training), [`models/segformer`](https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/segformer) (SegFormer training)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Dataset preparation"]
  node_dataset_build["Dataset building<br/>[build_dataset.py]"]
  node_dataset_split["Dataset splitting"]
  node_dataset_checks["Dataset verification<br/>[verify_dataset.py]"]
end

subgraph group_input["Data loading"]
  node_torch_dataset["Torch dataset<br/>[pytorch_dataset.py]"]
  node_root_dataset["Root dataset<br/>[pytorch_dataset.py]"]
  node_augment["Dataset augmentation"]
end

subgraph group_models["Segmentation models"]
  node_deeplab_model["DeepLabV3+"]
  node_deeplab_cbam["DeepLab CBAM"]
  node_deeplab_training["DeepLab training"]
  node_unet_model["U-Net model"]
  node_unet_training["U-Net training"]
  node_segformer_model["SegFormer model<br/>[segformer_model.py]"]
  node_segformer_training["SegFormer training"]
  node_segformer_variants["SegFormer variants"]
  node_deeplab_variants["DeepLab variants"]
end

subgraph group_results["Evaluation and analysis"]
  node_segformer_evaluation["SegFormer evaluation"]
  node_variant_evaluation["Variant evaluation"]
  node_analysis["Model analysis"]
  node_dataset_analysis["Dataset analysis"]
end

node_researcher(("Researcher"))

node_researcher -.->|"prepares data"| node_dataset_build
node_deeplab_training -->|"loads samples"| node_torch_dataset
node_deeplab_training -->|"creates model"| node_deeplab_cbam
node_unet_training -->|"loads samples"| node_root_dataset
node_unet_training -->|"creates model"| node_unet_model
node_segformer_training -.->|"uses dataset"| node_torch_dataset
node_segformer_training -.->|"uses model"| node_segformer_model
node_deeplab_variants -.->|"uses dataset"| node_torch_dataset
node_deeplab_variants -.->|"uses model"| node_deeplab_model
node_segformer_variants -.->|"uses dataset"| node_torch_dataset
node_segformer_variants -.->|"uses model"| node_segformer_model
node_segformer_evaluation -->|"loads test data"| node_root_dataset
node_variant_evaluation -->|"uses dataset"| node_torch_dataset
node_researcher -.->|"compares results"| node_analysis

click node_dataset_build "https://github.com/deepti0611/spacenet-off-nadir/blob/main/data_preparation/build_dataset.py"
click node_dataset_split "https://github.com/deepti0611/spacenet-off-nadir/blob/main/data_preparation/make_final_split.py"
click node_dataset_checks "https://github.com/deepti0611/spacenet-off-nadir/blob/main/checks/verify_dataset.py"
click node_torch_dataset "https://github.com/deepti0611/spacenet-off-nadir/blob/main/ml_pipeline/pytorch_dataset.py"
click node_root_dataset "https://github.com/deepti0611/spacenet-off-nadir/blob/main/pytorch_dataset.py"
click node_augment "https://github.com/deepti0611/spacenet-off-nadir/tree/main/ml_pipeline"
click node_deeplab_model "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_resnet50/deeplabv3plus_resnet50_model.py"
click node_deeplab_cbam "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_cbam_bce_dice/deeplabv3plus_cbam_model.py"
click node_deeplab_training "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py"
click node_unet_model "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/unet_resnet34/unet_resnet34_model.py"
click node_unet_training "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/unet_resnet34/train_unet_resnet34.py"
click node_segformer_model "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer/segformer_model.py"
click node_segformer_training "https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/segformer"
click node_segformer_variants "https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/segformer_cbam_bce_dice"
click node_deeplab_variants "https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/deeplabv3plus_resnet50_partial"
click node_segformer_evaluation "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer/evaluate_segformer.py"
click node_variant_evaluation "https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer_perspective/evaluate_segformer_perspective.py"
click node_analysis "https://github.com/deepti0611/spacenet-off-nadir/blob/main/analysis/segformer/generate_all_comparison_plots.py"
click node_dataset_analysis "https://github.com/deepti0611/spacenet-off-nadir/blob/main/analysis/analysis/building_area_analysis.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_dataset_build,node_dataset_split,node_dataset_checks toneBlue
class node_torch_dataset,node_root_dataset,node_augment toneAmber
class node_deeplab_model,node_deeplab_cbam,node_deeplab_training,node_unet_model,node_unet_training,node_segformer_model,node_segformer_training,node_segformer_variants,node_deeplab_variants toneMint
class node_segformer_evaluation,node_variant_evaluation,node_analysis,node_dataset_analysis toneRose
class node_researcher toneIndigo
```

## Components

### Dataset preparation

- [Dataset building](https://github.com/deepti0611/spacenet-off-nadir/blob/main/data_preparation/build_dataset.py): `data_preparation/build_dataset.py`
- [Dataset splitting](https://github.com/deepti0611/spacenet-off-nadir/blob/main/data_preparation/make_final_split.py): `data_preparation/make_final_split.py`
- [Dataset verification](https://github.com/deepti0611/spacenet-off-nadir/blob/main/checks/verify_dataset.py): `checks/verify_dataset.py`

### Data loading

- [Torch dataset](https://github.com/deepti0611/spacenet-off-nadir/blob/main/ml_pipeline/pytorch_dataset.py): `ml_pipeline/pytorch_dataset.py`
- [Root dataset](https://github.com/deepti0611/spacenet-off-nadir/blob/main/pytorch_dataset.py): `pytorch_dataset.py`
- [Dataset augmentation](https://github.com/deepti0611/spacenet-off-nadir/tree/main/ml_pipeline): `ml_pipeline`

### Segmentation models

- [DeepLabV3+](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_resnet50/deeplabv3plus_resnet50_model.py): `models/deeplabv3plus_resnet50/deeplabv3plus_resnet50_model.py`
- [DeepLab CBAM](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_cbam_bce_dice/deeplabv3plus_cbam_model.py): `models/deeplabv3plus_cbam_bce_dice/deeplabv3plus_cbam_model.py`
- [DeepLab training](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py): `models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py`
- [U-Net model](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/unet_resnet34/unet_resnet34_model.py): `models/unet_resnet34/unet_resnet34_model.py`
- [U-Net training](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/unet_resnet34/train_unet_resnet34.py): `models/unet_resnet34/train_unet_resnet34.py`
- [SegFormer model](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer/segformer_model.py): `models/segformer/segformer_model.py`
- [SegFormer training](https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/segformer): `models/segformer`
- [SegFormer variants](https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/segformer_cbam_bce_dice): `models/segformer_cbam_bce_dice`
- [DeepLab variants](https://github.com/deepti0611/spacenet-off-nadir/tree/main/models/deeplabv3plus_resnet50_partial): `models/deeplabv3plus_resnet50_partial`

### Evaluation and analysis

- [SegFormer evaluation](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer/evaluate_segformer.py): `models/segformer/evaluate_segformer.py`
- [Variant evaluation](https://github.com/deepti0611/spacenet-off-nadir/blob/main/models/segformer_perspective/evaluate_segformer_perspective.py): `models/segformer_perspective/evaluate_segformer_perspective.py`
- [Model analysis](https://github.com/deepti0611/spacenet-off-nadir/blob/main/analysis/segformer/generate_all_comparison_plots.py): `analysis/segformer/generate_all_comparison_plots.py`
- [Dataset analysis](https://github.com/deepti0611/spacenet-off-nadir/blob/main/analysis/analysis/building_area_analysis.py): `analysis/analysis/building_area_analysis.py`

### Other components

- **Researcher**

## Connections

- Researcher → Dataset building: prepares data
- DeepLab training → Torch dataset: loads samples (evidence: `models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py`)
- DeepLab training → DeepLab CBAM: creates model (evidence: `models/deeplabv3plus_cbam_focal/train_deeplabv3plus_cbam_focal.py`)
- U-Net training → Root dataset: loads samples (evidence: `models/unet_resnet34/train_unet_resnet34.py`)
- U-Net training → U-Net model: creates model (evidence: `models/unet_resnet34/train_unet_resnet34.py`)
- SegFormer training → Torch dataset: uses dataset
- SegFormer training → SegFormer model: uses model
- DeepLab variants → Torch dataset: uses dataset (evidence: `models/deeplabv3plus_resnet50_partial/deeplabv3plus_resnet50_partial_train.py`)
- DeepLab variants → DeepLabV3+: uses model (evidence: `models/deeplabv3plus_resnet50_partial/deeplabv3plus_resnet50_partial_train.py`)
- SegFormer variants → Torch dataset: uses dataset
- SegFormer variants → SegFormer model: uses model
- SegFormer evaluation → Root dataset: loads test data (evidence: `models/segformer/evaluate_segformer.py`)
- Variant evaluation → Torch dataset: uses dataset (evidence: `models/segformer_perspective/evaluate_segformer_perspective.py`)
- Researcher → Model analysis: compares results
