search

XiaoyuShi97 / VideoFlow

Official implementation of ICCV2023 VideoFlow: Exploiting Temporal Cues for Multi-frame Optical Flow Estimation
260 stars 29 forks source link

Generalization error on KITTI #16

Closed DQiaole closed 1 year ago

DQiaole commented 1 year ago

Hi, thanks for sharing your impressive work. I have downloaded the MOF_things.pth and changed _CN.model of configs/multiframes_sintel_submission.py accordingly. But after running command python -u evaluate_MOFNet.py --dataset=kitti, I get Validation KITTI: 4.284974, 15.392534, which is much worse than the results reported in the paper. How can I get the generalization results on KITTI(train) of Table 1 in the paper?

XiaoyuShi97 commented 1 year ago

Hi, thanks for your interest in our work. The Twins encoder is sensitive to resolution in the things stage, so we follow FlowFormer to train another model on resolution of 288*960 with noise augmentation. I will update that model soon.

XiaoyuShi97 commented 1 year ago

Hi. I have updated the 288*960 version. Please have a try.

DQiaole commented 1 year ago

Hi, thanks for sharing the model. I can now get similar results as yours.