DDAD Self-supervised Monocular Image Depth Estimation Dataset

DDAD stands for Dense Depth for Autonomous Driving, which is an urban driving dataset with higher challenges and accurate depth evaluation.

The dataset includes information from six calibrated cameras, as well as a high-resolution long-range LiDAR sensor to produce dense ground truth depth estimates up to 250 meters. The DDAD dataset contains multiple challenging and diverse urban environments in Japan and the United States, with a total of 12,650 anonymized samples collected.