Overhead imagery research data set — an annotated data library & tools to aid in the development of computer vision algorithms

When failures occur in machine object recognition algorithms, researchers may have limited information on the root causes of the failure. For example, did the algorithm fail to detect a target due to occlusion, shadow, contrast, or some other known computer vision shortcoming? The Overhead Imagery Research Data Set (OIRDS) project will help advance the state of the art in image processing and computer vision by providing an open-access, annotated overhead imagery library that will allow researchers to break down algorithm performance by image and target attributes. The OIRDS project has produced a data set with almost 1,000 labeled images suitable for developing automated vehicle detection algorithms. These images contain approximately 1,800 labeled targets. For each target, the OIRDS provides over 30 annotations and over 60 statistics that describe the target within the context of the image.