Transform Depth Maps into Stunning Point Clouds with Open3D and Python

Поділитися
Вставка
  • Опубліковано 8 лют 2025
  • Inside my school and program, I teach you my system to become an AI engineer or freelancer. Life-time access, personal help by me and I will show you exactly how I went from below average student to making $250/hr. Join the High Earner AI Career Program here 👉 www.nicolai-ni... (PRICES WILL INCREASE SOON)
    You will also get access to all the technical courses inside the program, also the ones I plan to make in the future! Check out the technical courses below 👇
    _____________________________________________________________
    In this video 📝 we are going to take a look at how to Create Our Own Point Clouds from Depth Maps in Open3D with Python. We will go over a couple of examples where we create point clouds from depth images together with the corresponding color image. We will use a monocular camera together with deep learning to estimate a depth map that will be used to generate out own point clouds.
    If you enjoyed this video, be sure to press the 👍 button so that I know what content you guys like to see.
    _____________________________________________________________
    🛠️ Freelance Work: www.nicolai-ni...
    _____________________________________________________________
    💻💰🛠️ High Earner AI Career Program: www.nicolai-ni...
    ⚙️ Real-world AI Technical Courses: (www.nicos-scho...)
    📗 OpenCV GPU in Python: www.nicos-scho...
    📕 YOLOv7 Object Detection: www.nicos-scho...
    📒 Transformer & Segmentation: www.nicos-scho...
    📙 YOLOv8 Object Tracking: www.nicos-scho...
    📘 Research Paper Implementation: www.nicos-scho...
    📔 CustomGPT: www.nicos-scho...
    _____________________________________________________________
    📞 Connect with Me:
    🌳 linktr.ee/nico...
    🌍 My Website: www.nicolai-ni...
    🤖 GitHub: github.com/nic...
    👉 LinkedIn: / nicolaiai
    🐦 X/Twitter: / nielsencv_ai
    🌆 Instagram: / nicolaihoeirup
    _____________________________________________________________
    🎮 My Gear (Affiliate links):
    💻 Laptop: amzn.to/49LJkTW
    🖥️ Desktop PC:
    NVIDIA RTX 4090 24GB: amzn.to/3Uc7yAM
    Intel I9-14900K: amzn.to/3W4Z5Cb
    Motherboard: amzn.to/4aR6wBC
    32GB RAM: amzn.to/3Jt2XVR
    🖥️ Monitor: amzn.to/4aLP8hh
    🖱️ Mouse: amzn.to/3W501GH
    ⌨️ Keyboard: amzn.to/3xUGz5b
    🎙️ Microphone: amzn.to/3w1F1WK
    📷 Camera: amzn.to/4b4Ryr9
    _____________________________________________________________
    Tags:
    #Open3D #PointClouds #ComputerVision

КОМЕНТАРІ • 40

  • @NicolaiAI
    @NicolaiAI  8 місяців тому

    Join My AI Career Program
    www.nicolai-nielsen.com/aicareer
    Enroll in My School and Technical Courses
    www.nicos-school.com

  • @damianhancock4035
    @damianhancock4035 3 роки тому +1

    Keep it up man. Loving the content

    • @NicolaiAI
      @NicolaiAI  3 роки тому +1

      Thanks for watching! Really appreciate it

  • @rajahaseeb8418
    @rajahaseeb8418 2 роки тому +6

    Hi. Thank you for your tutorials!
    I just wanted to ask a quick question.
    Suppose we have frames and we can get depth maps using any model. Then how can we generate point clouds for an entire scene using Open3D. I know how we can do it for one frame, but I am not sure how to accumulate all the frames in the right order and generate a whole map like SLAM.

  • @NicolaiAI
    @NicolaiAI  Рік тому +2

    📘Enroll in Research Paper Implementation Course: nicolai-nielsen-s-school.teachable.com/p/research-paper-implementation
    📗Enroll in Object Tracking Course:nicolai-nielsen-s-school.teachable.com/p/yolov8-object-tracking-course
    📕Enroll in YOLOv7 Course:nicolai-nielsen-s-school.teachable.com/p/yolov7-custom-object-detection-with-deployment

  • @FalconSmart
    @FalconSmart 3 роки тому +1

    Appreciate your work and sharing!!!!

  • @shivalukka6660
    @shivalukka6660 3 роки тому +1

    Hey man, are you planning to do a tutorial on ICP next? That would really help in my project. Keep up the good work!

    • @NicolaiAI
      @NicolaiAI  3 роки тому

      Yes that will be in one of the upcoming videos. Both ICP and global registration

    • @NicolaiAI
      @NicolaiAI  3 роки тому

      Thank you very much btw!

  • @prashantkramadhari3724
    @prashantkramadhari3724 3 роки тому +1

    Nice videos..Do you plan to use any deep learning approach to learn point cloud representation from depth map..? Perhaps even depth map

    • @NicolaiAI
      @NicolaiAI  3 роки тому +1

      Thank you very much! This video is basically deep learning to create point clouds. I have another with with point cloud classification and NN here on the channel. We are going to combine deep learning and point clouds way more im future videos! Stay tuned

  • @ahmetpehlivan5494
    @ahmetpehlivan5494 3 роки тому +1

    I assume you take your Q matrix from stereorectify method from previous videos, right?

    • @NicolaiAI
      @NicolaiAI  3 роки тому

      Yes that's correct since we need the perspective matrix to map our points in the opencv function. We could implement our own function that only uses monocular information and reproject the points out in that way

  • @ahmetpehlivan5494
    @ahmetpehlivan5494 3 роки тому +1

    great work! Can you make a video of creating STL files of these point clouds?

    • @NicolaiAI
      @NicolaiAI  3 роки тому

      Thank you very much! If u have a ply file u can actually just save it with Open3D as a STL file.

    • @ahmetpehlivan5494
      @ahmetpehlivan5494 3 роки тому

      @@NicolaiAI thank you 👍

  • @sharoseali708
    @sharoseali708 3 роки тому

    hi thanks for your nice work! Can you tell where to get onnx model that is used to calculate depth.. Thanks

  • @godkiller4020
    @godkiller4020 Рік тому

    Hey bro can u make a video on using tsdf integration to convert videos to depth images to 3d volumetric models

  • @SpOOnySk
    @SpOOnySk 3 роки тому +2

    Hello, I was trying to create depth map but I cant find the DNN model that you are using, is there any way you can upload it on your github?

    • @NicolaiAI
      @NicolaiAI  3 роки тому +2

      Forgot to link to the model in the description, here u go: github.com/isl-org/MiDaS/releases/tag/v2_1

    • @SpOOnySk
      @SpOOnySk 3 роки тому +1

      @@NicolaiAI Thank you so much, your tutorials helps me a lot with my diploma paper.

    • @NicolaiAI
      @NicolaiAI  3 роки тому

      Thanks for watching! Really glad that it can help u

  • @8eck
    @8eck 2 роки тому

    What is width=640 and height=480 of camera intrinsics? That's pixel resolution of your webcam?

  • @ahmetpehlivan5494
    @ahmetpehlivan5494 3 роки тому +1

    hello I couldnt find monoCalibrateCamera altough searching through al the files?

    • @NicolaiAI
      @NicolaiAI  3 роки тому +1

      It's just in the camera calibration python script on GitHub

    • @ahmetpehlivan5494
      @ahmetpehlivan5494 3 роки тому

      @@NicolaiAI bro if you are not the best i don't know anything 😅 I'll be around don't kick me out of the channel 😅✋

  • @farahzakiyahrahmanti8213
    @farahzakiyahrahmanti8213 2 роки тому

    Very nice tutorial, but how about transform 3D point cloud to 2D dept image?

  • @vineethakoduri7896
    @vineethakoduri7896 2 роки тому +1

    Great Video. It's really helpful for me.
    When I am trying with my color and depth images, I am getting the error like this "Invoked with: [PointCloud with 921600 points.]; kwargs: window_name='Open3D', width=1280, height=720, zoom=0.5", I tried on google colab. Can you please help me with this error.

    • @souvikdatta5686
      @souvikdatta5686 2 роки тому

      Hi were you able to solve it, since I encountered the same Error?

    • @vineethakoduri7896
      @vineethakoduri7896 2 роки тому

      @@souvikdatta5686 check the input size of you color and depth images and then see the rgbd size

    • @souvikdatta5686
      @souvikdatta5686 2 роки тому

      @@vineethakoduri7896 I just checked, my Colour and Depth Image are of size 600x400. But after that?
      Is it the number of channels?
      Since my original 2 images have 3 channels and RGBD Image has only 1

    • @vineethakoduri7896
      @vineethakoduri7896 2 роки тому

      @@souvikdatta5686 that right.you might get the points as 600x400x3= 720000.Dont try on colab, try locally

    • @souvikdatta5686
      @souvikdatta5686 2 роки тому

      @@vineethakoduri7896 How did you convert the number of channels from 3 to 1?

  • @souvikdatta5686
    @souvikdatta5686 2 роки тому

    I am facing an draw_geometries(): incompatible function arguments error on execution

    • @souvikdatta5686
      @souvikdatta5686 2 роки тому +1

      Answering my own question -> Just remove the zoom parameter from the function. The code will work just fine

  • @ghizlanekhattari8804
    @ghizlanekhattari8804 2 роки тому

    how i can convert image (png,gpg) to 3D without using deep learning with this libariry open3D ???????