datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
remote-sensing-sft-data
RSCoVLM: Co-Training Vision Language Models for Remote Sensing Multi-task Learning
Qingyun Li*
Shuran Ma*
Junwei Luo*
Yi Yu*
Yue Zhou
Fengxiang Wang
Xudong Lu
Xiaoxing Wang
Xin He
Yushi Chen
Xue Yang
If you find our work helpful, please consider giving us a ⭐!
ArXiv Paper: https://arxiv.org/abs/2511.21272
Published Paper: https://www.mdpi.com/2072-4292/18/2/222… See the full description on the dataset page: https://huggingface.co/datasets/Qingyun/remote-sensing-sft-data.remote_sensing_VQA_multilingual
Remote Sensing VQA — Multilingual
A multilingual counterfactual MCQ dataset built from remote sensing / satellite imagery.
Each row contains a satellite image, two captions (original vs counterfactual), and a multiple-choice question probing whether a VLM follows the image or the misleading text.
Languages
Language
Code
Rows
English
en
50
Hindi
hi
50
Urdu
ur
50
Telugu
te
50
Bahasa Indonesia
id
50
Columns
Column
Type… See the full description on the dataset page: https://huggingface.co/datasets/apart-global-south-hack/remote_sensing_VQA_multilingual.remote-sensing-VQA-benchmark
Adapting Multimodal Large Language Models to Domains via Post-Training (EMNLP 2025)
This repos contains the remote sensing visual instruction tasks for evaluating MLLMs in our paper: On Domain-Specific Post-Training for Multimodal Large Language Models.
The main project page is: Adapt-MLLM-to-Domains
1. Download Data
You can load datasets using the datasets library:
from datasets import load_dataset
# Choose the task name from the list of available tasks
task_name… See the full description on the dataset page: https://huggingface.co/datasets/AdaptLLM/remote-sensing-VQA-benchmark.remote-sensing-sample-library-manager
Remote Sensing Sample Library Manager
This repository contains a local desktop tool and its associated sample data for building remote-sensing element datasets from class-organized imagery and vector labels.
Contents
software/
dataset_filter_desktop_app/ # Tkinter desktop tool
data/
huanghekou_202508/ # full generated Huanghekou element dataset
huanghekou_202508_selected/ # selected subset generated by the tool… See the full description on the dataset page: https://huggingface.co/datasets/cuibinge/remote-sensing-sample-library-manager.remote_sensing_dbremote-sensing-change-detection
Remote Sensing Change Detection Dataset
For Chinese documentation, please see README_zh.md
Dataset Description
A specialized dataset for remote sensing change detection research, containing complete image processing pipeline and annotation information. This dataset includes 24 groups of registered and aligned remote sensing image samples, with each group containing 5 different types of image files and corresponding annotation files.
Dataset Features
Data… See the full description on the dataset page: https://huggingface.co/datasets/Mercyiris/remote-sensing-change-detection.remote-sensing-test
Remote Sensing Test Datasets
IC: Image Classification, OD: Object Detection, RC: Region Captioning, GD: Grounding Description, Captions: Captions, VQA:Visual Question Answering
ship_classification_2024_2025
Description
Collection of remotely sensed scenes of cropped boxes for the ship
detection project for Remote Sensing in Grenoble-INP ENSE3 2024/2025.
The boxes that contain or not are contained in the folders ship and
no_ship, respectively.
The related project is available at:
https://gricad-gitlab.univ-grenoble-alpes.fr/piconed/remote-sensing-projects-archive
in the folder projects/2024_2025/08_ship_detection
Credits
Hugo Fournier… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/ship_classification_2024_2025.pollution_2024_2025
Description
Dataset of Sentinel-5P acquisitions with the TROPOMI sensor.
Data was collected for the project of pollution monitoring
for the course of remote sensing in ENSE3, Grenoble-INP during
the academic year of 2024/2025.
The dataset contains:
co_canada_2025: CO acquisition in Canada from 01/05/2025 to 08/05/2025
co_paris_2025: CO acquisition in Paris from 10/05/2025 to 14/05/2025
co_random: extra CO acquisitions
no2_paris_2022: NO_2 acquisitions in Paris from 01/05/2022 to… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/pollution_2024_2025.remote-sensing-VQAremote-sensing-visual-instructions
Adapting Multimodal Large Language Models to Domains via Post-Training (EMNLP 2025)
This repos contains the remote-sensing visual instructions for post-training MLLMs in our paper: On Domain-Specific Post-Training for Multimodal Large Language Models.
The main project page is: Adapt-MLLM-to-Domains
Data Information
Using our visual instruction synthesizer, we generate visual instruction tasks based on the image-caption pairs from NWPU-Captions, RSICD, RSITMD… See the full description on the dataset page: https://huggingface.co/datasets/AdaptLLM/remote-sensing-visual-instructions.ship_detection_roboflow_2024_2025
Description
Collection of remotely sensed scenes with ships for the project of
ship detection for Remote Sensing in Grenoble-INP ENSE3 2024/2025.
The images were originally collected on Roboflow:
https://roboflow.com
The related project is available at:
https://gricad-gitlab.univ-grenoble-alpes.fr/piconed/remote-sensing-projects-archive
in the folder projects/2024_2025/08_ship_detection
Credits
Hugo Fournier (Hugo.Fournier2@grenoble-inp.org)Lilian Haemmerer… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/ship_detection_roboflow_2024_2025.forest-plot-analysis
Description
This dataset contains the information related to seven forest field plots located in the Belledonne massif, Northern Alps, France.
The data is separated into:
LIDAR: a collection of 24 features computed from a 3D point cloud;
Hyperspectral imagery, collected with the Hysper VNIR 1600 sensor;
Ground truth: An annotation of the 13 tree species;
Further details are given in the report contained in the repository.
Credits
If you use this data, please cite:… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/forest-plot-analysis.ship-detection
Description
This dataset is a subset of the one provided for the Ship Detection Competion organized by Data-Driven Science, originally available here:
https://huggingface.co/datasets/datadrivenscience/ship-detection
This subset contains 10 images of ships and a JSON with the coordinates associated to the bounding box of the available ships.
Credits
Data-Driven Science
Low-Altitude-Drone-Remote-Sensing-Dataset
Low-Altitude-Drone-Remote-Sensing-Dataset
Dataset Description
The Low-Altitude-Drone-Remote-Sensing-Dataset is a high-resolution drone image dataset collected at a consistent low altitude over residential neighborhoods in coastal urban areas, Galveston, Texas. The dataset is designed for remote sensing image compression, image restoration, and downstream visual understanding tasks.
Compared with many existing UAV datasets captured under highly diverse viewpoints and… See the full description on the dataset page: https://huggingface.co/datasets/yuminghan12123/Low-Altitude-Drone-Remote-Sensing-Dataset.multimodal-fusion-remote-sensing-datagrenoble_zanzibar
Description
This dataset contains data acquired by both the Landsat and Sentinel-2 platforms relative to a time series between the year 2020 to 2024 relative to the areas of Grenoble, France and the region of Zanzibar.
In particular, the dataset contains:
animation: A small animation relative to the evolution of certain quality indices
landsat
clipped: Clipped regions to the AOI (area of interest)
full_patches: Raw data
lst: Land Surface Temperature
multispectral: Full band… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/grenoble_zanzibar.deforestation-amazon
Description
This dataset contains a time series of acquisitions taken with the Sentinel 2 satellite platform at different times of the year whose area of interest is in the Amazon forest.
Main project
https://gricad-gitlab.univ-grenoble-alpes.fr/dallamum/remote-sensing-projects/
HLS_Remote_Sensinggrenoble_urban_2024_2025
Description
Collection of remotely sensed image of the scene of Grenoble for the project of
urban classification for Remote Sensing in Grenoble-INP ENSE3 2024/2025.
The related project is available at:
https://gricad-gitlab.univ-grenoble-alpes.fr/piconed/remote-sensing-projects-archive
in the folder projects/2024_2025/03_urban
Credits
Celian Charrin (Celian.Charrin@grenoble-inp.org)
Alexandre Jolly (Alexandre.Jolly@grenoble-inp.org)
Morgan Santalucia… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/grenoble_urban_2024_2025.remotesensingremote_sensing_2018_weedmapThe WeedMap dataset is a comprehensive collection of multispectral images captured from sugar beet fields in Eschikon, Switzerland, and Rheinbach, Germany, using quadrotor UAVs equipped with RedEdge-M and Sequoia multispectral cameras.
Spanning over five months, it comprises 129 directories with 18,746 image files. The dataset is divided into Orthomosaic and Tiles folders, featuring orthomosaic maps and their segmented tiles, respectively.
Ground truth annotations are provided, detailing classifications such as background, crop, and weed in both color and indexed formats. This dataset, the largest publicly available for sugar beet fields with pixel-level ground truth, spans a total area of 16,554 square meters.
It offers a detailed representation of the agricultural landscape, including a ground sample distance of about 1cm, facilitating high precision in weed detection research. This rich dataset supports the development of advanced deep learning models for semantic segmentation in precision agriculture, enhancing weed management practices.RemoteSensingCorpusopenearthmap-paris
Description
This dataset is a subset of the OpenEarthMap Land Cover Mapping Data, originally available here:
https://open-earth-map.org/overview_oem.html
The original dataset was subsampled to just contain the region of paris.
The dataset is composed of RGB images and associated annotated components.
Credits
Full dataset available on Zenodo:
https://zenodo.org/records/7223446
@inproceedings{xia2023openearthmap,
title={Openearthmap: A benchmark dataset for global… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/openearthmap-paris.corine-land-cover
Description
Dataset taken by the database of CORINE for land cover for the region of Rhone Alpes in France.
The dataset provides a set of classes associated to the terrain.
tropomi-pollution
Description
This is a file for the tracking of the pollution taken with the TROPOMI sensors, which tracks gases and aerosol relative to the air quality.
The data is in NETCDF format and relative to the 27/09/2020.
Main repository
https://gricad-gitlab.univ-grenoble-alpes.fr/dallamum/remote-sensing-projects
modis-snow-coverage
Description
This is a set of images taken by the MODIS sensor in the year 2013.
For each folder the dataset contain:
Modimlab_date_reproj2.tif: Fused image. 7 bands at 250m
Cloud_date_reproj2.tif: Cloud mask (1: cloud, 0: no cloud)
Wholesnow_ date_reproj2.tif: Fractional snow cover obtained by MODImLAB
Ndsi_date_reproj.tif: Fractional snow fraction based on NDSI
Spot_degrade_date.mat: Fractional snow cover based on Spot acquisitions at the same date and resampled to 250m… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/modis-snow-coverage.meteo-greener
Description
This dataset contains acquisitions taken by the meteo station at Greener, acquired by the team Shiny.
The website is available:
https://g2elab-shiny.g2elab.grenoble-inp.fr/meteo-greener/#dashboard
In particular the dataset is composed of images taken hourly by an emispheric camera sensor.
The data in the examples folder refers to acquisitins in winter, summer and rain conditions:
19-01-02: January 2nd, 2019
19-07-01: August 1st, 2019
rain: January 2nd, 2019
Two CSV… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/meteo-greener.srtm-grenoble
Description
This dataset contains a series of digital elevation data from the
SRTM (Shuttle Radar Topography Mission) dataset.This file contains a Digital Elevation Model (DEM) of the Earth's surface,
giving you the elevation in meters at regularly spaced intervals
(90m grid here). The values are integers representing
meters above sea level.
For this dataset, the targeted area is around Grenoble, France.
Download instructions
Visit… See the full description on the dataset page: https://huggingface.co/datasets/remote-sensing-ense3-grenoble-inp/srtm-grenoble.Land-Use-Cover-Remote-Sensing-Dataset
Land Use Cover Remote Sensing Dataset
The current agricultural sector faces rapid depletion of land resources and environmental changes. Especially against the backdrop of global climate change, efficient monitoring of land use has become particularly important. Existing remote sensing datasets are mostly focused on urban areas with insufficient attention to agricultural land coverage, leading to a lack of data support for agricultural planning and environmental assessment. This… See the full description on the dataset page: https://huggingface.co/datasets/Mobiusi/Land-Use-Cover-Remote-Sensing-Dataset.