evaluate the CAD2016 dataset

[qzhai]

Hi,

Thank you for sharing the exciting work.

I found that the evaluation doesn't work for CAD2016 (lines 168 to 184 in dataloaders/video_list.py). The image and the GT sequences are not of the same length, and I found that every 5 frames sampling is executed on the image sequences to evaluate the CAD2016 dataset. Could you please update the correct script you used during the evaluation?

Thanks.