This is the official website of the OODT dataset.

⭐ The dataset application is very simple and requires only the following two steps:

• Please fill in this application form.
• Please send your completed application form to this E-mail address:rs_devotee@163.com.
  When we receive your application, we will reply as soon as possible. Thank you for your support!

OODT is the first dataset of oriented objects for object detection and tracking in satellite vedio datasets (SVDs), in which each object is labeled as a rectangular box with rotating orientation. It consists of the single-object and multi-object tracking datasets.
The single-object tracking (SOT) dataset consists of 74 sequences, including 30 cars, 17 planes, 17 ships and 10 trains with 22610 frames in total.

The multi-object tracking (MOT) datasets is tentatively labeled with 8628 vehicles in 41 frames of five SVDs .

The raw satellite video data is collected from various sources, such as the JiLin-1 satellite constellation ^[1], SkySat satellite constellation^[2], object tracking contests ^[3][4][5], and some other publicly available videos ^[6][7].

SVD1 and SVD2 are collected from JiLin-1 satellite constellation ^[8]. SVD3 is collected from IRIS camera on the International Space Station^[3] .SVD4 and SVD5 are collected from SkySat satellite constellation^[9][10].

If you have any questions, please contact rs_devotee@163.com.

• The single object tracking section of the OODT dataset has been organized and counted in detail. You can view them in the OOTB.

• In order to further improve the richness of multi-object detection data and better simulate the actual application scenarios, we annotated the LMOD dataset with the HBB (Horizontal Bounding Box) annotation method from the perspectives of large-scale and multi-class features.