routerset

routerset is a materialized multi-label remote sensing dataset assembled from the current phi2FM downstream sources.

What is kept in this repo

• multilabel_dataset/manifest.jsonl: canonical record manifest
• multilabel_dataset/images/: materialized .npy samples
• multilabel_dataset/label_vocab.json: label vocabulary
• multilabel_dataset/summary.json: dataset build summary
• multilabel_dataset/materialization_summary.json: materialization results
• multilabel_dataset/plots/: dataset analysis graphs

The repo is intentionally minimal. Redundant root-level metadata copies and training-ready derivative exports were removed.

Current snapshot

• Records: 15,512
• Materialized samples: 15,512 .npy files
• Included datasets: fire, burned_area, anomaly_detection, worldfloods, lc, roads
• Missing dataset: building
• Label cardinality: min 0, max 5, mean 0.8083

Per-dataset totals

• fire: 1,600
• burned_area: 6,660
• anomaly_detection: 36
• worldfloods: 2,353
• lc: 63
• roads: 4,800

Status counts

• positive: 10,514
• explicit_negative: 2,777
• below_threshold: 2,221

Most frequent labels

• cloud: 4,159
• road_present: 2,702
• burned_area: 2,087
• water: 1,710
• land: 1,448

Analysis plots

The dataset graphs are stored in multilabel_dataset/plots/:

• coverage_distributions.png
• dataset_label_heatmap.png
• dataset_split_counts.png
• label_cooccurrence.png
• label_frequency.png
• status_and_cardinality.png
• plot_summary.json

Notes

• Arrays are stored as .npy materializations.
• The dataset was built from the current local phi2FM downstream sources and then uploaded to this Hugging Face dataset repository.