Pretrained weights on the kitti dataset for Adv attacks

[Jianyuan Wang]

Hi authors,

I am trying to evaluate the optical flow networks' performance against attacks. I was wondering if the MaskFlownet model pretrained on the kitti dataset can be available?

Best Regards

[Yilun (Simon) Sheng]

Hi, thanks for your interest!

Yes, of course, but it takes some time to rename the weight prefixes and I am having midterm right now. I will try to make it available this weekends.

[Jianyuan Wang]

Hi Simon,

May I ask how is it going with the renaming process?

Hope all the best.

[Yilun (Simon) Sheng]

Hi, I am sorry that I was quite busy with the school things recently.

We have only done validation on the KITTI dataset for the MaskFlownet-S. I am wondering if that will be what you are looking for.

[Jianyuan Wang]

Hi Simon,

Thanks for your quick reply. I just commented to check if the model is available. It would be great if the reported MaskFlownet model can be released.

Good luck with your school things.

[Yilun (Simon) Sheng]

Hi Jianyuan,

I am not sure what do you mean by the reported MaskFlownet model. Anyway, I have just uploaded a pretrained weights for MaskFlownet, called 8caNov12-1532, which has been fine-tuned on the whole KITTI dataset.

Please run something like python main.py MaskFlownet.yaml -g 0 -c 8caNov12 --valid --resize 576,1280 to do validation on it with inference resize (576, 1280).

[Jianyuan Wang]

Thanks a lot, Simon