SIGNet: Semantic Instance Aided Unsupervised 3D Geometry Perception

Benchmark Model Rank Results
monocular-depth-estimation-on-kitti-eigen-splitSIGNet#65absolute relative error: 0.133