VME: A Satellite Imagery Dataset and Benchmark for Detecting Vehicles in the Middle East and Beyond
Paper • 2505.22353 • Published • 1
images list | annotations list | categories list |
|---|---|---|
[{"height":512,"width":512,"id":1,"file_name":"VME0738.png"},{"height":512,"width":512,"id":2,"file_(...TRUNCATED) | [{"iscrowd":0,"image_id":1,"bbox":[59.0,348.0,8.0,9.0],"segmentation":[],"category_id":0,"id":1,"are(...TRUNCATED) | [{"id":0,"name":"Car","supercategory":"Car"},{"id":1,"name":"Bus","supercategory":"Bus"},{"id":2,"na(...TRUNCATED) |
- Paper:: arXiv:2505.22353
VME Dataset
Vehicles in the Middle East (VME) is a satellite imagery dataset built for vehicle detection in the Middle East.
VME images (satellite_images folder) are under CC BY-NC-ND 4.0 license and available for download on Zenodo repository, we also release VME ground-truth annotations and the scripts associated for constructing CDSI (annotations_HBB, annotations_OBB, CDSI_construction_scripts) under CC BY 4.0 license which are available on Zenodo repository and here.
This repository consists of:
- VME_annotations, which has:
- annotations_OBB: It holds TXT files in YOLO format with Oriented Bounding Box (OBB) annotations. Each annotation file is named after the corresponding image name.
- annotations_HBB: This component contains HBB annotation files in JSON file formatted in MS-COCO format defined by four values in pixels (x_min, y_min, width, height) of training, validation, and test splits.
- satellite_images: This folder consists of VME images of size 512x512 in PNG format. (The VME images are available in the Zenodo repository)
- Downloads last month
- 159