NimbleD: Enhancing Self-supervised Monocular Depth Estimation with Pseudo-labels and Large-scale Video Pre-training

Benchmark Model Rank Results
monocular-depth-estimation-on-kitti-eigen-1NimbleD-LiteMono-8M#7absolute relative error: 0.092RMSE: 4.194RMSE log: 0.165
monocular-depth-estimation-on-kitti-eigen-1NimbleD-LiteMono#16absolute relative error: 0.096RMSE: 4.304RMSE log: 0.171
monocular-depth-estimation-on-kitti-eigen-1Nimbled-SwiftDepth#17absolute relative error: 0.096RMSE: 4.333RMSE log: 0.171
monocular-depth-estimation-on-kitti-eigen-1Nimbled-MD2-R50#18absolute relative error: 0.097RMSE: 4.377RMSE log: 0.172
monocular-depth-estimation-on-kitti-eigen-1Nimbled-SwiftDepth-S#20absolute relative error: 0.098RMSE: 4.401RMSE log: 0.174
monocular-depth-estimation-on-kitti-eigen-1NimbleD-LiteMono-S#24absolute relative error: 0.099RMSE: 4.370RMSE log: 0.172
monocular-depth-estimation-on-kitti-eigen-1Nimbled-MD2-R18#25absolute relative error: 0.100RMSE: 4.440RMSE log: 0.175