Correction to “Background Subtraction Using Multiscale Fully Convolutional Network”

In the above paper
[1]
, the authors did not cite a reference that was crucial to the paper. Please note that the following citation
[2]
should have been cited in the following paragraph (page 16014): “From the output binary masks we can see that there is a great contrast between the foreground objects and their back-grounds, and in most cases the foreground objects are uniform connected areas, which means that in the input frames, there should also have a big difference of the features between the foreground and the background
[2]
.” and “So in order to extract this kind of contrast information, we add a contrast layer
[2]
after the feature layer
X_i
.”