absolute relative error: 0.128
Papers with Code · CC BY-SA 4.0
Depth Estimation
Depth Estimation is the task of measuring the distance of each pixel relative to the camera. Depth is extracted from either monocular (single) or stereo (multiple views of a scene) images. Traditional methods use multi-view geometry to find the relationship between the images. Newer methods can directly estimate depth by minimizing the regression loss, or by learning to generate a novel view from a sequence. The most popular benchmarks are KITTI and NYUv2. Models are typically evaluated according to a RMS metric.
<span class="description-source">Source: DIODE: A Dense Indoor and Outdoor DEpth Dataset </span>
논문
absolute relative error: 0.114RMSE: 4.706RMSE log: 0.191Delta < 1.25: 0.873
absolute relative error: 0.12
absolute relative error: 0.140
absolute relative error: 0.135
absolute relative error: 0.106
absolute relative error: 0.112
absolute relative error: 0.091
absolute relative error: 0.104RMSE: 4.494RMSE log: 0.181Delta < 1.25: 0.888
absolute relative error: 0.149
absolute relative error: 0.093
absolute relative error: 0.113
absolute relative error: 0.096
absolute relative error: 0.126
absolute relative error: 0.149Delta < 1.25: 0.784
absolute relative error: 0.096
absolute relative error: 0.099RMSE: 3.304RMSE log: 0.138Delta < 1.25: 0.916
absolute relative error: 0.14
absolute relative error: 0.133
absolute relative error: 0.096
absolute relative error: 0.096
absolute relative error: 0.133
absolute relative error: 0.110
absolute relative error: 0.094
absolute relative error: 0.097
absolute relative error: 0.113RMSE: 4.812RMSE log: 0.191Delta < 1.25: 0.877
absolute relative error: 0.118
absolute relative error: 0.193