DETR - End to end object detection with transformers (ECCV2020)

Поделиться
HTML-код
  • Опубликовано: 16 дек 2024

Комментарии • 25

  • @fire_nakamura
    @fire_nakamura 14 дней назад +1

    I'm fascinated by you and your team members' craft, with tweaks on loss, ideas of encodings and sufficient amount of data, applications will be huge. I would love to learn and explore those possibilities, Isn’t there anyway to be a part of your team or contribute to any related projects?

  • @kvnptl4400
    @kvnptl4400 6 месяцев назад

    A very nice presentation with clear visualizations and easy-to-understand explanations! Great Work!!🌟🌟🌟🌟🌟
    Smooth animations 👌

  • @QuintinMassey
    @QuintinMassey 2 года назад +3

    Outstanding work. I’m also very interested in the, arguably more difficult, small object detection problem.

  • @syedabdul8509
    @syedabdul8509 3 года назад +7

    Excellent Explanation.
    But I want to know the most important thing in this video,
    How did you create those cool animations like @1:58-@2:20 and @8:00-@8:05

  • @MarioHari
    @MarioHari 4 года назад +2

    Nice work!
    A small correction to what you said: "Semantic segmentation labels each pixel in the whole image. It is not restricted to only pixels in the background".

    • @nicolascarion3111
      @nicolascarion3111  4 года назад +5

      You're right, my statement is imprecise. I meant that semantic annotations of foreground classes are not used in the panoptic task.

    • @MarioHari
      @MarioHari 4 года назад

      @@nicolascarion3111 merci infiniment :)

    • @ujjalkrdutta7854
      @ujjalkrdutta7854 2 года назад

      @@nicolascarion3111 Can we then say that: "Panoptic Segmentation= Instance Segmentation+Semantic Segmentation minus annotations of foreground classes" ?

  • @Ramakrishnan-bq9is
    @Ramakrishnan-bq9is 3 года назад +1

    Thanks for sharing!
    Could you please explain what you mean by full differentiable and how other methods might not be fully differentiable?

    • @goldenshale
      @goldenshale 2 года назад

      This is an end to end neural network defined by functions which all have derivatives. In the R-CNN family of algorithms you have one procedure that produces a bunch of region proposals, then you crop out these regions and feed them to a classifier, and then you run another algorithm to prune out overlapping and low confidence predictions. Since there are multiple steps that have logical rather than mathematical implementations, you can't take derivatives all the way through to back propagate information through the whole system.

  • @morancium
    @morancium 26 дней назад

    WoW thankyou for your contribution!

  • @Nino234mff
    @Nino234mff 3 года назад

    Thank you for the great work and the presentation!

    • @kaceangelo132
      @kaceangelo132 3 года назад

      i realize it is quite off topic but do anyone know of a good website to watch new movies online ?

    • @bakercain265
      @bakercain265 3 года назад

      @Kace Angelo try Flixzone. Just google for it =)

  • @chandrahasp6697
    @chandrahasp6697 Год назад

    Really good work!

  • @ujjalkrdutta7854
    @ujjalkrdutta7854 2 года назад

    Elegant explanation. liked it

  • @rohinim7707
    @rohinim7707 4 года назад +1

    Amazing! What was the main motivation behind using a sequence model for an object detection?

    • @redjammie8342
      @redjammie8342 4 года назад

      It is not a sequence model. It was successfully used for sequences, but it's not a sequence model by definition.

  • @ZobeirRaisi
    @ZobeirRaisi 4 года назад +1

    What this mean?: "since the transformer is a permutation
    equivalent some extra care is required to retain
    the 2d structure of the image."

    • @nicolascarion3111
      @nicolascarion3111  4 года назад +7

      The transformer isn't aware of the 2D structure of the image, because 1) we flatten it and 2) permuting the inputs of a transformer simply permutes its outputs (permutation equivariance). That's why we add 2D positional encodings. This is similar to what is done in NLP, to retain the order of the sentence.

    • @ZobeirRaisi
      @ZobeirRaisi 4 года назад +1

      @@nicolascarion3111 Thanks for your explanation. I have another question: Right now DETR because of rectangle bboxes of COCO-dataset produces rectangle-bboxes outputs, if we had polygon bboxes (8 points), which parts of the architecture must be modified to output a polygon shape bboxes?

    • @nicolascarion3111
      @nicolascarion3111  4 года назад +4

      @@ZobeirRaisi Well you need to modify the regression head as well as the loss and matching function (GiOU may not make sense anymore, so you'll likely have to stick to L1). For this kind of questions, it's best to open an issue on our github. Thanks!