Problem about unzipping MUVFET-II dataset

[KuangJuiHsu]

Hi,

This work is good, and I'm interested in your work.

However, some error messages occur when I unzip MUVFET-II dataset, and they display as the follows:

"E:\MUVFET-II.zip: Checksum error in E:\MUVFET-II\database\MUVFET-II\202.avi. The file is corrupt
E:\MUVFET-II.zip: Checksum error in E:\MUVFET-II\database\MUVFET-II\155.avi. The file is corrupt
E:\MUVFET-II.zip: The archive is corrupt"

I use WinRAR to unzip this dataset on window 10.

Could you check if MUVFET-II is okay, please?

Thank you very much.

[tonysy]

Sorry for this problem, I have uploaded the MUVFET-II zip file again. I think the prolem caused by the bad network state(we can only connect the google server with VPN, which is very slow.). Please try the new file and contact me if you have any problem.

[KuangJuiHsu]

Thank you for providing your data.

When reading the video frames of MUVFET-II, the warning message occurs as the follows:

"Warning: The end of file was reached before the requested frames were read completely. Frames 1 through 575 were returned. "

I download the videos via BaiduYun, and I find the same warning message, too. Is this correct or not?

In addition, I have some questions about your work.

1. When I run your demo code, the generated masks are not so accurate. For example, I see the first frame of 024.mp4, and I find that the softmask doesn't cover the face. Is it correct?
   

2. Is the generated softmask, i.e. "fixmaps", in the demo code the ground truth for evaluating your results in Tabel 2 and visualization in Figure 8 and 9?

3. In Table 2, I don't understand how to calculate the results of GT for NSS and CC in Table 2? In addition, could you provide the evaluation code?

4. What is the definition of the saliency weight of each face, \w_{n,t}?