YOLO 9000 Object Detection #7

แชร์
ฝัง
  • เผยแพร่เมื่อ 24 พ.ย. 2024

ความคิดเห็น • 46

  • @darioturchi1958
    @darioturchi1958 6 ปีที่แล้ว +21

    I really liked the "living thing" label

    • @litpath3633
      @litpath3633 6 ปีที่แล้ว

      I was hoping for horseless carriage.

    • @Skaxarrat
      @Skaxarrat 6 ปีที่แล้ว +2

      "organism"

    • @DingXiaoke
      @DingXiaoke 6 ปีที่แล้ว

      Me too lol

  • @HiTechTopTIPs
    @HiTechTopTIPs 6 ปีที่แล้ว +20

    please provide specification of your processor, GPUs, RAM etc

  • @xuyuancn
    @xuyuancn 7 ปีที่แล้ว +2

    were imags downsampled in preprocessing or?

  • @tamirtap
    @tamirtap 4 ปีที่แล้ว +1

    Just to make sure, the input image is resized to the corresponding backbone your used (i.e. YOLO9000 input image is 544x544x3) right? or you've trained YOLO on HD images?

    • @KarolMajek
      @KarolMajek  4 ปีที่แล้ว +1

      It's resized.
      For larger input resolution check YOLOv3 (608), and faster R-CNN NasNet (1200)

  • @jam63112
    @jam63112 5 ปีที่แล้ว +2

    Containers everywhere

  • @1d10tcannotmakeusername
    @1d10tcannotmakeusername 6 ปีที่แล้ว +1

    What does it classify a Furry as?

  • @Qizot
    @Qizot 6 ปีที่แล้ว

    Does it really operate on 4k or it transforms it to lower resolution and then apply detection?

  • @muhammedsert9482
    @muhammedsert9482 6 ปีที่แล้ว

    Can image processing be followed by insects like creatures?

  • @ghumakkad_baccha
    @ghumakkad_baccha 6 ปีที่แล้ว

    hello sir. i m trying to install opencv and contrib-opencv. but i could not install it. i have done face ditection in my laptop with opencv. will you plz give me some help to install it on my raspberry pi 3

  • @pseudooduesp2805
    @pseudooduesp2805 6 ปีที่แล้ว

    super with large number and random trajectory of cars . good job .
    can i ask you some course or tutorial to learn made the same things ?
    thanks for advance

  • @shanakaj007
    @shanakaj007 5 ปีที่แล้ว

    Karol, I would suggest you to also save the raw videos and make a huge repository. I think your work could be very big because we'd need data to fuel the deep learning..(like open soruce datasets)

    • @KarolMajek
      @KarolMajek  5 ปีที่แล้ว

      I already shared one video
      archive.org/details/0002201705192

    • @leonelkellan5289
      @leonelkellan5289 3 ปีที่แล้ว

      instaBlaster.

  • @duguder
    @duguder 6 ปีที่แล้ว

    Can't find a yolo app on IOS?

  • @Juanp082413
    @Juanp082413 5 ปีที่แล้ว

    Hello Man I`m looking for a program to count persons, motorcycles and vehicles for my job. Tell me if U can help me. Thanks

    • @KarolMajek
      @KarolMajek  5 ปีที่แล้ว

      smth like this: th-cam.com/video/m2CqoZuBOeo/w-d-xo.html

  • @HerrLehmann0xcc
    @HerrLehmann0xcc 6 ปีที่แล้ว +2

    does someone know what the colors mean?

    • @xinzhou3744
      @xinzhou3744 6 ปีที่แล้ว

      It means category I think, cars are all blue, etc

    • @HerrLehmann0xcc
      @HerrLehmann0xcc 6 ปีที่แล้ว

      ok, thanks. I thought maybe it could be something like "expected to find" vs "unexpected". It's obviously designed to work on single images instead of video though.

    • @KarolMajek
      @KarolMajek  6 ปีที่แล้ว +4

      Yes it is meant to work on single image, knowledge from previous frames is ignored. After adding some tracking it should look a lot better!

    • @AlanDeRossett
      @AlanDeRossett 6 ปีที่แล้ว

      classifier's each trained set could be different. We use Red for urgent objects like recognizing a Weapon or as a Warning for alerts and lock downs/

  • @qzorn4440
    @qzorn4440 6 ปีที่แล้ว +1

    sweet program, what about deer anti-crash actions? thanks. ;*/

  • @bzqp2
    @bzqp2 6 ปีที่แล้ว +1

    Dude, it's just vanilla 9000. You need to train it on something...

    • @KarolMajek
      @KarolMajek  6 ปีที่แล้ว +1

      No, I don't need

  • @ManuelRobalinho
    @ManuelRobalinho 6 ปีที่แล้ว

    Very good!

  • @jackspragnow5071
    @jackspragnow5071 5 ปีที่แล้ว +1

    real life wall hack esp :D

  • @akd9977
    @akd9977 4 ปีที่แล้ว

    Superb. I suggest you how you did programmatically so that others can learn from you. Thanks

  • @aevrembidiones4478
    @aevrembidiones4478 4 ปีที่แล้ว

    Hi Good day, I am a business student. I have a feasibility study for Oil and energy industry. can you help me by creating a software that counts a different type of vehicle such as van, car, truck and etc. I am willing to pay. thank you

  • @seedinnovationspvtltd.937
    @seedinnovationspvtltd.937 6 ปีที่แล้ว

    Do you have idea on modifying YOLO for detecting small objects, similar to those in satellite/aerial imagery? I found one great solution to this on: medium.com/@avanetten, they modified YOLO to YOLT(you only look twice and modified the pipeline, and since they didnt describe the workflow or any tutorials, i have no clue on how to implement it.
    Any help would be highly appreciated.

  • @tiendct
    @tiendct 6 ปีที่แล้ว +1

    Please, tell me about: hardware (CPU, VGA, v..) need for realtime detection the same video

    • @KarolMajek
      @KarolMajek  6 ปีที่แล้ว +2

      Laptop, i7 6700HQ, GTX980M
      This is not realtime. I just uploaded new version of YOLO COCO video with FPS

    • @AlanDeRossett
      @AlanDeRossett 6 ปีที่แล้ว

      pjreddie.com/darknet/hardware-guide/

  • @evansyomu2879
    @evansyomu2879 5 ปีที่แล้ว

    specs of processor, and GPU, thats all i need

    • @KarolMajek
      @KarolMajek  5 ปีที่แล้ว

      Laptop i7 with 8 threads, GTX980m (bought in 2016)

  • @nmb4951
    @nmb4951 5 ปีที่แล้ว

    Dicshonal Pregident JAPAN Messege nec pcv-eddr domein mmc pcv.