Total3DUnderstanding: Joint Layout, Object Pose and Mesh Reconstruction for Indoor Scenes from a Single Image

Benchmark Model Rank Results
3d-shape-reconstruction-on-pix3dMGN#2CD: 0.0836EMD: N/AIoU: N/A
monocular-3d-object-detection-on-sun-rgb-dTotal3D joint#3AP@0.15 (NYU-37): 14.28AP@0.15 (10 / NYU-37): 26.38
monocular-3d-object-detection-on-sun-rgb-dTotal3D w/o. joint#4AP@0.15 (NYU-37): 13.25AP@0.15 (10 / NYU-37): 23.32
room-layout-estimation-on-sun-rgb-dTotal3D joint#3IoU: 59.2Camera Pitch: 3.15Camera Roll: 2.09
room-layout-estimation-on-sun-rgb-dTotal w/o. joint#4IoU: 57.6Camera Pitch: 3.68Camera Roll: 2.59