Questions about video guided machine translation task

[PingYu-iris]

Hi there,

Recently, I read your paper, which is quite interesting. However, after reading, I come across one problem about this video-guided machine translation task.

In your example in Figure 1, your example shows video-guided method could improve the machine translation results. However, I tested the current natural machine translation tool (such as the mBART or google translation), I found their translation result is near perfect even without video signals. The question is: have your compare your video-guided machine translation results with neural machine translation methods? Is that helpful for using the video signals?

Best,
Ping