Real-Time Semantic Background Subtraction - ICIP 2020 - #3

Поділитися
Вставка
  • Опубліковано 3 гру 2024

КОМЕНТАРІ • 14

  • @lumiere2598
    @lumiere2598 3 роки тому +1

    Thank you for the wonderful presentation!

  • @mcehd1323
    @mcehd1323 4 роки тому +1

    Interesting topic and clear explanations!

  • @boxiaoma7160
    @boxiaoma7160 4 роки тому +1

    Cool idea and impressive presentation!

  • @WSam1993
    @WSam1993 3 роки тому

    This is great! I am wondering if this could work on greyscale videos? I believe the semantic segmentation wouldn't work, right?

    • @ACADResearch
      @ACADResearch  3 роки тому

      Hi, thanks for your comment and sorry about the late reply. This is a great question! Actually, this method and its components are not limited to RGB images. In fact, there are also thermal images, which are in the greyscale format, in the evaluation dataset, CDNet2014. However, you should probably look if PSPNet works well on your own greyscale images as the performance of the overall method will depend on that. If not, one possibility is to simply replace PSPNet with a segmentation network you know works well on your particular data.

  • @mancio4369
    @mancio4369 3 роки тому

    wow nice video bro!

  • @swatidewangan6110
    @swatidewangan6110 4 роки тому

    How can I segment a plant disease spot in an image? How the masks are created?

    • @ACADResearch
      @ACADResearch  3 роки тому

      Hi! The semantic segmentation masks are created with PSPNet, then these masks are combined with the change detection masks of Vibe to produce the final background subtraction masks. However, I believe that this method might not be suited for plant disease spot detection, as it is more focused on motion.

  • @yagnicchandra1489
    @yagnicchandra1489 3 роки тому

    nice video