DICEPTION: A Generalist Diffusion Model for Visual Perceptual Tasks

Benchmark Model Rank Results
monocular-depth-estimation-on-diode-relative-depth-alignedDICEPTION Single (zero-shot)#1AbsRel: 0.283
monocular-depth-estimation-on-diode-relative-depth-alignedDICEPTION (zero-shot)#6delta1: 0.722
monocular-depth-estimation-on-kitti-relative-depth-alignedDICEPTION Single (zero-shot)#1AbsRel: 0.064delta1: 0.952
monocular-depth-estimation-on-eth3d-relative-depth-alignedDICEPTION (zero-shot)#3AbsRel: 0.050delta1: 0.975
monocular-depth-estimation-on-scannet-relative-depth-alignedDICEPTION (zero-shot)#12AbsRel: 0.072delta1: 0.944
monocular-depth-estimation-on-nyu-depth-v2DICEPTION (zero-shot)#18absolute relative error: 0.061Delta < 1.25: 0.960
human-pose-keypoint-estimation-on-coco-val2017DICEPTION–AP: 57.8