DN Detr (Denoising Detr for object detection)

Поделиться
HTML-код
  • Опубликовано: 17 янв 2025

Комментарии • 24

  • @A.El-Taher
    @A.El-Taher 10 месяцев назад +1

    Finally, bro ❤🎉

  • @A.El-Taher
    @A.El-Taher 10 месяцев назад +1

    After this video i will try Dn detr

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Happy to hear that

  • @A.El-Taher
    @A.El-Taher 10 месяцев назад +1

    I realy support you 😊

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Warms my heart!

  • @davidro00
    @davidro00 10 месяцев назад +1

    Such a good explanation! Thank you :)
    Why do the authors choose a fixed number of denoised boxes instead of using the number of actual ground truth boxes?
    So what happens if there are e.g. 15 objects in the ground truth and there are only 10 denoising queries?

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Great question!
      Total number of denoising queries is gt_objects * number_of_groups. So for 5 groups, 15 gt objects - you will have 75 denoising queries. For 2 gt objects, you will have just 10 denoising queries

    • @davidro00
      @davidro00 10 месяцев назад +1

      @@makgaiduk ah i didnt get that, thanks!

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      I am making a next video about reading the source code for dn detr. Helps clear out ambiguities like these )@@davidro00

    • @davidro00
      @davidro00 10 месяцев назад

      @@makgaiduk looking forward💪🏼

    • @vipuljrml2778
      @vipuljrml2778 10 месяцев назад

      @@makgaiduk Hi! Thank you for a great video! Can you explain what "group" means here?

  • @A.El-Taher
    @A.El-Taher 10 месяцев назад +1

    Keep going 💪

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Thanks! I will!

  • @A.El-Taher
    @A.El-Taher 10 месяцев назад +1

    I was trying Dab detr in the morning

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Keep exploring the topic! Hopefully my videos could help

  • @makgaiduk
    @makgaiduk  9 месяцев назад

    Check out my next video: Reading DN Detr source code ruclips.net/video/6DDJjoLtsyA/видео.html

  • @shafamw
    @shafamw 10 месяцев назад

    would you please review rt-detr and DINO detr?

    • @makgaiduk
      @makgaiduk  10 месяцев назад

      Most certainly will! I am doing DINO after SWIN, which will be my next video. It's also one of the DINO components. You can expect DINO in around 3 weeks. Rt detr will take a bit more time, but I will definitely do it!