Peter Wonka Research
Peter Wonka Research
  • 100
  • 75 155
3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models
3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models; Biao Zhang, Jiapeng Tang, Matthias Niessner, Peter Wonka; ACM Siggraph / ACM Transactions on Graphics 2023
Keywords: Neural Fields; 3D Generative Modeling; Generative Diffusion Models; Shape Autoencoding; 3D Data Structures
arxiv.org/abs/2301.11445
มุมมอง: 1 239

วีดีโอ

LocalBins: Improving Depth Estimation by Learning Local Distributions
มุมมอง 547ปีที่แล้ว
LocalBins: Improving Depth Estimation by Learning Local Distributions SF Bhat, I Alhashim, P Wonka ECCV 2022 Short Paper Presentation
DLFVC - 60 - Neural Architecture Search
มุมมอง 3973 ปีที่แล้ว
Introduction to Neural Architecture Search; NAS; Deep Learning;
DLFVC - 50 - Point Clouds - Part 4 / 4
มุมมอง 3233 ปีที่แล้ว
Deep Learning; Applications of Point Cloud Processing; Classification; Segmentation; Instance Segmentation; Detection; Normal Estimation;
DLFVC - 50 - Point Clouds - Part 3 / X
มุมมอง 2703 ปีที่แล้ว
Deep Learning; Point Clouds; PointNet , DGCNN; KPConv;
DLFVC - 50 - Point Clouds Part 2 / X
มุมมอง 1963 ปีที่แล้ว
PointNet CVPR 2017; details of the original paper; deep learning;
DLFVC - 50 - Point Clouds Part 1 / X
มุมมอง 3153 ปีที่แล้ว
Introduction to Point Clouds; Deep Learning
Tutorial: Integer Programming for Visual Computing - Part 5 / 5
มุมมอง 973 ปีที่แล้ว
Mixed Integer Quadratic Programming (MIQP); Quadratic Assignment Problem; Mixed Integer Quadtractically Constrained Programming (MIQCP);
DLVC - 41 - Generative Adversarial Networks - Part 5 / 5
มุมมอง 1143 ปีที่แล้ว
Generative Adversarial Networks; Cycle Consistency; Various Networks
Tutorial: Integer Programming for Visual Computing - Part 3 / 5
มุมมอง 673 ปีที่แล้ว
Examples of Integer Programming involving Graphs; Minimum Spanning Tree; Traveling Salesman; Tourist Trip Planning;
Tutorial: Integer Programming for Visual Computing - Part 2 / 5
มุมมอง 813 ปีที่แล้ว
Examples of Linear Integer Programming Formulations; Knapsack Problem; Linear Assignment; Map Labeling; Tiling; Camera Placement; Shape Matching; Tourist Map Layouts;
Tutorial: Integer Programming for Visual Computing - Part 1 / 5
มุมมอง 3143 ปีที่แล้ว
Introduction to Optimization and Integer Programming
DLVC - 41 - Generative Adversarial Networks - Part 4 / X
มุมมอง 1213 ปีที่แล้ว
GAN Example from the original paper; DC GAN
DLVC - 41 - Generative Adversarial Networks - Part 3 / X
มุมมอง 893 ปีที่แล้ว
Explanation of the original GAN equation and general GAN setup; Deep Learning
DLFVC - 41 - Generative Adversarial Networks Part 1 / X
มุมมอง 1643 ปีที่แล้ว
DLFVC - 41 - Generative Adversarial Networks Part 1 / X
DLFVC - 41 - Generative Adversarial Networks Part 2 / X
มุมมอง 753 ปีที่แล้ว
DLFVC - 41 - Generative Adversarial Networks Part 2 / X
DLFVC - 41 - Generative Adversarial Networks Part 1 / X broken video
มุมมอง 1543 ปีที่แล้ว
DLFVC - 41 - Generative Adversarial Networks Part 1 / X broken video
DLVC - 33 - Per Pixel Output: Segmentation, Depth Estimation, ... - Part 2 / X
มุมมอง 933 ปีที่แล้ว
DLVC - 33 - Per Pixel Output: Segmentation, Depth Estimation, ... - Part 2 / X
DLVC - 33 - Per Pixel Output: Segmentation, Depth Estimation, ... - Part 1 / X
มุมมอง 1423 ปีที่แล้ว
DLVC - 33 - Per Pixel Output: Segmentation, Depth Estimation, ... - Part 1 / X
DLFVC - 18 - Network Analysis - Part 3 / 3
มุมมอง 603 ปีที่แล้ว
DLFVC - 18 - Network Analysis - Part 3 / 3
DLFVC - 18 - Network Analysis - Part 2 / 3
มุมมอง 463 ปีที่แล้ว
DLFVC - 18 - Network Analysis - Part 2 / 3
DLFVC - 18 - Network Analysis - Part 1 / 3
มุมมอง 733 ปีที่แล้ว
DLFVC - 18 - Network Analysis - Part 1 / 3
DLFVC - D02 - Visualization using Tensorboard
มุมมอง 933 ปีที่แล้ว
DLFVC - D02 - Visualization using Tensorboard
DLFVC - 17 - Visualization
มุมมอง 1713 ปีที่แล้ว
DLFVC - 17 - Visualization
DLFVC - 16 - Classification Architectures Part 2/2
มุมมอง 1073 ปีที่แล้ว
DLFVC - 16 - Classification Architectures Part 2/2
DLFVC - 16 - Classification Architectures Part 1 / 2
มุมมอง 1083 ปีที่แล้ว
DLFVC - 16 - Classification Architectures Part 1 / 2
DLFVC - 31 - Style Transfer
มุมมอง 2453 ปีที่แล้ว
DLFVC - 31 - Style Transfer
DLFVC - 30 - Texture Synthesis
มุมมอง 3093 ปีที่แล้ว
DLFVC - 30 - Texture Synthesis
DLFVC - 14 - Part 2/3 - Introduction, SGD, Momentum
มุมมอง 1693 ปีที่แล้ว
DLFVC - 14 - Part 2/3 - Introduction, SGD, Momentum
DLFVC - 14 - Part 3/3 - RMSPropo, ADAM, RAdam, Lookahead, Novograd
มุมมอง 3123 ปีที่แล้ว
DLFVC - 14 - Part 3/3 - RMSPropo, ADAM, RAdam, Lookahead, Novograd

ความคิดเห็น

  • @gogopet
    @gogopet หลายเดือนก่อน

    I would like to commission a project from you, how do I contact you?

  • @OsamaMustafa-DTU
    @OsamaMustafa-DTU 4 หลายเดือนก่อน

    Thank you for your invaluable advice. I'm enthusiastic about the prospect of working under your mentorship to further develop these abilities.

  • @asimshaikh5350
    @asimshaikh5350 5 หลายเดือนก่อน

    Part 2 please

  • @user-sc1os1qo3x
    @user-sc1os1qo3x 6 หลายเดือนก่อน

    May I ask if there are any written records or blogs about this content?

  • @raghavamorusupalli7557
    @raghavamorusupalli7557 11 หลายเดือนก่อน

    Sir, thank you for your incredible efforts in the preparation of the contents and for presenting them so elegantly!

  • @raghavamorusupalli7557
    @raghavamorusupalli7557 11 หลายเดือนก่อน

    Wonderful stuff sir

  • @flake22
    @flake22 ปีที่แล้ว

    Great video and very helpful. That was the missing link in data preprocessing for me. One follow up question about multi dimensional tensors. For example, I have a timeseries of sensor readings that I want to feed into a Transformer Encoder model. I want to treat all sensor readings per timestep as a feature vector. The dataset has then the shape (#samples, sequence_length, features) Since the sensor readings are in different magnitudes, I absolutely need to normalize it. Should I normalize it individually over one sample (meaning over one sequence) to keep the temporal relationship, or should I normalize it over the whole dataset, like in your 2d example? Second question: In some NLP tasks the word embeddings are also normalized but with L2 normalization such that the embedding vector is scaled to unit length (over last axis). You also presented a few more normalizations like min-max, max-abs and robust. In which scenario is it best to use what normalization? Can one combine multiple normalizations, like minmax and mean/var? Which normalization technique would you recommend, given the description of my dataset above? Sorry for the long question, I couldn't find good information about normalization. Greetings.

  • @gogopet
    @gogopet ปีที่แล้ว

    Are you developing this project still?

  • @harilalmn
    @harilalmn ปีที่แล้ว

    Hi.. Very interesting... I wish to contact you to know further about the approach and the algorithm...

  • @Metrawolf
    @Metrawolf ปีที่แล้ว

    Whatever became of this project after this SIGGRAPH presentation?

  • @surjayshivakota3939
    @surjayshivakota3939 2 ปีที่แล้ว

    can i ouder some art

  • @amursaqibpal587
    @amursaqibpal587 2 ปีที่แล้ว

    Great lecture, thank you!

  • @chriss740
    @chriss740 2 ปีที่แล้ว

    I presume the 'arm' used is a general purpose robotic device, used for other purposes? Otherwise, this looks like an expensive hobby!

  • @adhammohamedsalaheldeen7988
    @adhammohamedsalaheldeen7988 2 ปีที่แล้ว

    thank you for this video, I was just wondering how for example the input is 256x256 and the output is 512x512?

  • @gogopet
    @gogopet 2 ปีที่แล้ว

    Is that softaware I would like to buy

    • @jeremylin8465
      @jeremylin8465 6 วันที่ผ่านมา

      Somebody re-made it github.com/ProbableTrain/MapGenerator/tree/master

    • @gogopet
      @gogopet 6 วันที่ผ่านมา

      ​@@jeremylin8465 wow wow Thank you thank you

  • @samahelsiedy
    @samahelsiedy 2 ปีที่แล้ว

    String art th-cam.com/video/KWeNLzpAPE0/w-d-xo.html

  • @94Husam
    @94Husam 2 ปีที่แล้ว

    Great explanations, thank you

  • @yanpogutsa
    @yanpogutsa 2 ปีที่แล้ว

    7 years passed no one had even tried to implement this workflow(

    • @mr.m2675
      @mr.m2675 11 หลายเดือนก่อน

      i guess Plasticity 3D did

  • @cyanmargh
    @cyanmargh 2 ปีที่แล้ว

    Hello. Is this channel still alive? I would like to contact you and discuss some of your old videos. I am wondering how you have implemented some of the functionality. Could you provide an email address for further correspondence, please?

  • @cyanmargh
    @cyanmargh 3 ปีที่แล้ว

    very underrated video

  • @kirillqes3062
    @kirillqes3062 3 ปีที่แล้ว

    how to make a portrait more realistic? please tell me.

  • @VNcarpenter
    @VNcarpenter 3 ปีที่แล้ว

    Maths 😞

  • @eduardolarrymarinsilva76
    @eduardolarrymarinsilva76 3 ปีที่แล้ว

    Would it be possible to implement a similar technique, but allowing the thread to wrap around the outside of several pins? I think that may reduce the artifacts of the image

    • @binyaminmanela3585
      @binyaminmanela3585 3 ปีที่แล้ว

      To do that, you would have to look atleast two steps ahead (or other wised the algorithm wouldn'tsee the advantageof such a move), which will multiply the time complexity.

    • @WillHirschUK
      @WillHirschUK 2 ปีที่แล้ว

      The difference between wrapping around the outside and doing short connections between adjacent pins is pretty negligible (in any case the area nearest the edge of the circle ends up completely filled in with thread) so provided the search algorithm can explore connecting enough of these together it will already be doing this.

    • @chriss740
      @chriss740 2 ปีที่แล้ว

      @@binyaminmanela3585 Isn't that what computers are good at ...? Also, couldn't you simply re-write the algorithm to allow 'free form' lines without the constraint that the end of one line must be the start of the next line? That way, the 'plan' would be for thousands of independent lines, and then the issue of wrapping thread around the outside of the pins would just be a matter of logistics - getting the thread to where it needs to be.

  • @openroomxyz
    @openroomxyz 3 ปีที่แล้ว

    If is Art than you should no destroy it at end, but put it in gallery or in your home.

  • @chris24hdez
    @chris24hdez 3 ปีที่แล้ว

    3:28 Nooooo!

  • @andrewworries4558
    @andrewworries4558 3 ปีที่แล้ว

    0:10 This private method earn 5000$ in 2 days codz2019.blogspot.com/2020/09/150-case-study.html

  • @sitiramadhani
    @sitiramadhani 3 ปีที่แล้ว

    I really hope there is an opportunity for joining your research group prof.

  • @sitiramadhani
    @sitiramadhani 3 ปีที่แล้ว

    awfull thank for your sharing Prof.

  • @fantashio
    @fantashio 3 ปีที่แล้ว

    Nice work! Let's grow together.

  • @kotvkvante22
    @kotvkvante22 3 ปีที่แล้ว

    sounds cool, but I didn't understand anything

  • @konstantinburlachenko2843
    @konstantinburlachenko2843 3 ปีที่แล้ว

    03:52 - overview of the program guide in cemse.kaust.edu.sa/cs/study#programguide 13:00 - definitions in the program guide 16:25 - non-regular courses 17:50 - a requirement for all students 24:40 - MS program requirement 28:45 - Ph.D. program requirement 31:00 - Ph.D. (course-based) qualification exams. Briefly: 3 eligible course from different bins/category (from different faculty)

  • @TheEngineerGuy
    @TheEngineerGuy 4 ปีที่แล้ว

    Is it going to be an online series?

  • @clydemarindo3177
    @clydemarindo3177 4 ปีที่แล้ว

    I know that this is a relatively old video, but where can I learn quadrangulations especially quadrangulating with arbitary points in each side.

  • @enriquejara699
    @enriquejara699 4 ปีที่แล้ว

    ah yes the good old windows xp

  • @liyunzhang4276
    @liyunzhang4276 4 ปีที่แล้ว

    it looks great, but have you calculate the accuracy of the reconstructed model compared with ground truth?

  • @better_dead_than_red
    @better_dead_than_red 4 ปีที่แล้ว

    Балин хочу уже на самоизоляцию с сосачерами в Битардск Че там куда балки нести?

  • @ajaykhedkar6497
    @ajaykhedkar6497 4 ปีที่แล้ว

    Hello sir, It's very nice video. could you tell me which software is this? please upload similar videos on optimization lattice dome structure and I will be very thankful if tell me some references for optimization. thank you...

  • @ep1taph31
    @ep1taph31 4 ปีที่แล้ว

    I hate when they rip it apart at the end

  • @mohamedalhammadi7840
    @mohamedalhammadi7840 5 ปีที่แล้ว

    Which software you used to transfare a photo to art

  • @meta3d562
    @meta3d562 5 ปีที่แล้ว

    Is it possible for you to share the used code? leonardoayache@hotmail.com

    • @randomrandom316
      @randomrandom316 3 ปีที่แล้ว

      If you are still on the look out and Matlab works for you then - th-cam.com/video/M1gXuKFspgY/w-d-xo.html&ab_channel=BartonDring This talks about the matlab code you might be looking out for.

  • @kayleezhu7328
    @kayleezhu7328 5 ปีที่แล้ว

    Thanks for the video! I am wondering how can I download the software you demonstrate in the video?

  • @austinrivera6421
    @austinrivera6421 6 ปีที่แล้ว

    Great work! Do you have MATLAB files you'd be willing to share to utilize / test the methods?

  • @jefersonromuloy
    @jefersonromuloy 6 ปีที่แล้ว

    Is it possible to get input data examples that were used in the paper?

  • @izmaelverhage
    @izmaelverhage 7 ปีที่แล้ว

    Is there a public code example? I'd like to do vector texture mapping in my pet project, maybe this will work well.

  • @colahuang1251
    @colahuang1251 8 ปีที่แล้ว

    Good job! However, the interaction is a little complex from the single view image. Could multi view images improve the accuracy and automation of the reconstruction?

  • @JunkerJames
    @JunkerJames 8 ปีที่แล้ว

    This is like magic!

  • @iceanimemusicvideo
    @iceanimemusicvideo 9 ปีที่แล้ว

    That's amazing... Hope I can do this someday too...

  • @HTTimChen
    @HTTimChen 9 ปีที่แล้ว

    great work!

  • @benoeilnoir
    @benoeilnoir 11 ปีที่แล้ว

    Very good job ! Do you share more info on a blog or something?

  • @TiagoTiagoT
    @TiagoTiagoT 12 ปีที่แล้ว

    I hope there are no legal obstacles (nor technical ones) for getting this on Blender...