ClearSightNet output color format


I’m trying to understand ClearSightNet’s output format. From what I saw in the sample video, the network outputs an RGB image for each input frame. Can you let me know the meaning of each channel?

From what I see, it seems like the the network outputs a visibility level per pixel and for the purpose of visualisation, red, green, blue pixel corresponds to heavily, moderately and lightly occluded (black means fully-visible). Or each channel is representing different class of occlusion/visibility (red is occlusion, green is visibility, etc…)?

Thanks a lot

Dear bphan,
It creates mask indicates image regions that are blurred, blocked, or clean. Red: blocked, Black: clean, Green: blurred. Please check for more details about the sample.

Hi Siva,

How about the blue channel? I ran it with one of the video and there’s a blue region in the output



Do you have any update on this?


Dear bphan,
I could reproduce you ask on my machine. Checking internally on this. I will update you once I hear back from internal team

Dear @bphan,
sorry for late update. the blue color indicates sky. We are updating the documentation to have this information.