Monodepth2

Monodepth2

Results on KITTI raw test set (Mono)

Backbone	Resolution	Sup	Train	Abs Rel.	Sq Rel.	RMSE	RMSElog	A1	Model
Res18	192x640	Mono	Reported	0.115	0.903	4.863	0.193	0.877	-
Res18	192x640	Mono	Official	0.115	0.909	4.866	0.193	0.877	Baidu/Google
Res18	192x640	Mono	Trained	0.113	0.858	4.753	0.190	0.879	Baidu/Google
Res18	320x1024	Mono	Reported	0.115	0.882	4.701	0.190	0.879	-
Res18	320x1024	Mono	Official	0.115	0.886	4.704	0.190	0.879	Baidu/Google
Res18	320x1024	Mono	Trained	0.109	0.797	4.533	0.184	0.887	Baidu/Google

Results on KITTI raw test set (Stereo)

Backbone	Resolution	Sup	Train	Abs Rel.	Sq Rel.	RMSE	RMSElog	A1	Model
Res18	192x640	Stereo	Reported	0.109	0.873	4.960	0.209	0.864	-
Res18	192x640	Stereo	Official	0.109	0.875	4.959	0.209	0.864	Baidu/Google
Res18	192x640	Stereo	Trained	0.108	0.858	4.819	0.202	0.866	Baidu/Google
Res18	320x1024	Stereo	Reported	0.107	0.849	4.764	0.201	0.874	-
Res18	320x1024	Stereo	Official	0.107	0.851	4.765	0.201	0.874	Baidu/Google
Res18	320x1024	Stereo	Trained	0.104	0.824	4.747	0.200	0.875	Baidu/Google

Results on Make3D test set (trained on KITTI)

Backbone	Resolution	Sup	PP	Train	Abs Rel.	Sq Rel.	RMSE	log10	Model
Res18	192x640	Stereo		Reported	0.322	3.589	7.417	0.163
Res18	192x640	Stereo		Official	0.321	3.379	7.254	0.163	Baidu/Google
Res18	192x640	Stereo		Trained	0.311	3.166	6.959	0.158	Baidu/Google

• Reported means that the results are reported in the paper.
• Official means that the results are predicted with the models got from their Official Implementations.
• Trained means that the results are predicted with the models trained with this repository.
• code for all the download links of pan Baidu is smde.