Datasetpublic

ALI-ezz/urban-curated-yolo26l

ALI-ezz/urban-curated-yolo26l 7-class urban remote sensing detection dataset in YOLO format. Aligned with DeepGeoAgent v5 PDF §3 (project1-detection.pdf): Classes (7) 0: building 1: rooftop_solar_panel 2: road_segment 3: construction_zone 4: green_urban_patch 5: water_body 6: parking_facility Splits (geographically separated per PDF §5) train/: DeepGlobe 2018 road extraction (Thailand/India) val/: NAIP aerial (USA) + synthesized urban features… See the full description on the dataset page: https://huggingface.co/datasets/ALI-ezz/urban-curated-yolo26l.

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes189downloads

Dataset Card

ALI-ezz/urban-curated-yolo26l

7-class urban remote sensing detection dataset in YOLO format. Aligned with DeepGeoAgent v5 PDF §3 (project1-detection.pdf):

Classes (7)

—0: building
—1: rooftopsolarpanel
—2: road_segment
—3: construction_zone
—4: greenurbanpatch
—5: water_body
—6: parking_facility

Splits (geographically separated per PDF §5)

—train/: DeepGlobe 2018 road extraction (Thailand/India)
—val/: NAIP aerial (USA) + synthesized urban features
—test/: APKLOT parking lots (Spain) + MS Egypt building shells

Format

Each split has images/*.jpg and labels/*.txt (YOLO cls cx cy w h normalized). Sidecar *.source files track the original dataset each tile came from.

Citation

DeepGeoAgent v5 — Kaggle notebook, 7-class urban object detection.