Skip to content

Dataset · IGNF

French agricultural satellite image time series

IGNF/PASTIS-HD

Benchmark for panoptic and semantic segmentation of agricultural parcels using satellite data.

What's inside

Contains 2,433 fragments of satellite imagery over France with pixel-level panoptic labeling (object index and semantic label). Each fragment is a multi-spectral Sentinel-2 image time series of variable length, also including radar series and very high-resolution imagery.

What you can do with it

  • Training crop segmentation and classification models using satellite imagery
  • Comparing time series analysis algorithms for Earth observation data
Tasks
Image classification, Image segmentation
Size
1K–10K rows
Downloads this month
24.5K
Likes
20

How to load it

from datasets import load_dataset

ds = load_dataset("IGNF/PASTIS-HD")