'Deep Learning/Papers2read'에 해당되는 글 119건

  1. 2020.12.07 안녕하세요 딥러닝 논문읽기 모임 입니다! 논문 발표자료 : https://www2.slideshare.net/taeseonryu/explai
  2. 2020.08.31 1주 1논문 리뷰 프로젝트 공유 안녕하세요. TFKR! 대학 여름방학 마지막 주를 맞아, 제가 이번 방학동안 진행한 1주 1논문 리뷰 프로젝트
  3. 2020.08.27 A novel neural network to generate high-resolution images For project and code/A
  4. 2020.08.05 # 질문 있습니다! 첨부한 파일은 성별 전환 애플리케이션(Application)으로 레오나르도 디카프리오(Leonardo DiCaprio)의
  5. 2020.07.29 COCO-WholeBody dataset is the first large-scale benchmark for whole-body pose es
  6. 2020.07.25 Latest from Stanford and Adobe Researchers: Inferring 3D human motion from video
  7. 2020.07.24 Latest from Max Planck researchers: State of the art in Shape and Pose Disentang
  8. 2020.07.24 From #ECCV2020: Reconstruct a morphable shape, texture, and viewpoint from an i
  9. 2020.07.23 Facebook released a Machine Learning algorithm (Multilevel Pixel Aligned Implici
  10. 2020.07.22 Latest from Microsoft researchers: High-quality video inpainting! For project a
  11. 2020.07.20 Latest from Baidu researchers: Automatic video inpainting algorithm that can rem
  12. 2020.07.17 Agile Human Behavior Imitation by humanoid models! For project and code/API/exp
  13. 2020.07.17 Better than Faceapp: State of the art in Facial Attribute Editing! For project
  14. 2020.07.08 Latest from Adobe and UC Berkeley researchers: State of the art in deep image ma
  15. 2020.07.08 Deep single image manipulation using conditional adversarial generators! For pro
  16. 2020.06.29 오늘 소개드릴 논문은 볼수록 여러 BIZ에 적용할 만한 사업 아이디어가 많이 떠올랐던 재미있는 논문이었습니다. [동영상1]은 드라마에서 캡쳐한
  17. 2020.06.17 A Neural Rendering Framework for Free-Viewpoint Relighting 중국 상하이에서 나온 뉴럴 렌더링 논
  18. 2020.06.17 Faces à la Carte: Text-to-Face Generation via Attribute Disentanglement 얼굴을 묘사한
  19. 2020.06.15 안녕하세요 FAIR에서 일하고 있는 주한별입니다. 이번에 CVPR oral에 발표될 PIFuHD라는 논문을 소개해드립니다. 이미지 한장으로 사람
  20. 2020.06.15 Latest from Microsoft researchers: Recovering the 3D geometry of human head from
  21. 2020.06.15 Interesting Research!!! S2IGAN — Speech-to-Image Generation via Adversarial Lea
  22. 2020.06.12 From SIGGRAPH 2020: Method reconstructs the geometry of complex 3D thin structur
  23. 2020.06.11 Reconstruct 3D human body shapes based on a sparse set of RGBD frames using a si
  24. 2020.06.09 Recapture your portrait photos with desired posture/view, figure, and clothing s
  25. 2020.06.04 Latest from Samsung researchers: State of the art in photo editing (Harmonizatio
  26. 2020.06.04 DeepFaceDrawing: Deep Generation of Face Images from Sketches Paper: http://geom
  27. 2020.06.03 היי חברים, אני מעביר הרצאות בנושא של מידול תלת מימדי מתמונה בעזרת רשתות ניורנים.
  28. 2020.06.03 Latest from Apple researchers: Deep learning approach for driving animated faces
  29. 2020.05.29 From Adobe researchers: State of the art in High-Resolution Image Inpainting For
  30. 2020.05.20 Adversarial Colorization of Icons Based on Structure and Color ConditionsAuthors: Tsai-Ho Sun, Chien-Hsun Lai, Sai-Keung Wong, and Yu-Shuen WangAbstract: We present a system to help #designers create icons that are widely used in banners, signboards, bi..

안녕하세요 딥러닝 논문읽기 모임 입니다!

논문 발표자료 : https://www2.slideshare.net/taeseonryu/explaining-knowledge-distillation
지금까지 업데이트 된 논문 리뷰 영상 : https://github.com/Lilcob/-DL_PaperReadingMeeting
딥논읽 채널 : https://www.youtube.com/channel/UCDULrK2OJsiDhFroa2Aj_LQ?view_as=subscriber

오늘 업데이트된 따끈따근한 논문 리뷰 영상은
'Explaining Knowledge Distillation by Quantifying the Knowledge'
라는 제목의 논문입니다 해당논문은 2020 cvpr에서 프로시딩된 논문입니다!
해당논문은 DNN의 인풋 이후 메인 태스크와 관련이 있는 이미지와 관련이 없는 이미지의 feature를 찾아내어 시각적으로 정량화하고 분석하여 knowledge distillation을 성공적으로 수행하는방법을 제시합니다! 또한 해당논문은 세가지의 가설을 제시하여 논문의 메소드에대한 설명을 보다 탄탄하게 뒷받침 하는대요,
펀디멘탈팀의 김동희님이 자세한 논문 리뷰를 도와주셨습니다!
오늘도 많은 관심 감사드립니다!
https://youtu.be/3q3iHhgKOY0

Posted by uniqueone
,

1주 1논문 리뷰 프로젝트 공유
안녕하세요. TFKR! 대학 여름방학 마지막 주를 맞아, 제가 이번 방학동안 진행한 1주 1논문 리뷰 프로젝트를 공유하고자 합니다!
여름방학 8주간 5편의 딥러닝 관련 논문을 리뷰하는 영상을 만들어 유튜브에 업로드하였습니다. 영상 시청보다 자료를 선호하시는 분들을 위한 PDF 자료도 준비되어 있습니다!
영상: https://www.youtube.com/playlist?list=PLDU04UBu69OG6mJ8m5osSxxQ0CFSej_e1
PDF: https://github.com/skyil7/paperReview
회고: https://skyil.tistory.com/93
논문 선정, 리뷰 과정 등에서 항상 TFKR 분들께 많은 도움을 받고 있습니다! 제 자료가 좋은 선순환을 만드는데 도움이 되었으면 좋겠네요😊
TFKR 여러분 남은 여름 건강하고 알차게 보내시길 바랍니다! 감사합니다~

Posted by uniqueone
,

A novel neural network to generate high-resolution images
For project and code/API/expert requests: https://catalyzex.com/paper/arxiv:2008.10399

Posted by uniqueone
,

# 질문 있습니다!

첨부한 파일은 성별 전환 애플리케이션(Application)으로 레오나르도 디카프리오(Leonardo DiCaprio)의 성별을 '여성'으로 전환시킨 영상입니다.

요는 어떤 GAN 모델을 사용하면 위와 같은 작업을 할 수 있을까요?

StarGAN Version 2 논문 1쪽에 나오는 그림 1을 보고 판단해보면 StarGAN Version 2로도 가능할까요?

Posted by uniqueone
,

COCO-WholeBody dataset is the first large-scale benchmark for whole-body pose estimation. It is an extension of COCO 2017 dataset with the same train/val split as COCO.

For project and code/API/expert requests: https://www.catalyzex.com/paper/arxiv:2007.11858

For each person, they annotate 4 types of bounding boxes (person box, face box, left-hand box, and right-hand box) and 133 keypoints (17 for body, 6 for feet, 68 for face and 42 for hands).

Get the free ML code finder browser extension:
Chrome https://bit.ly/code_finder_chrome
Firefox https://bit.ly/code_finder_firefox.

Posted by uniqueone
,

Latest from Stanford and Adobe Researchers: Inferring 3D human motion from video sequences that takes initial 2D and 3D pose estimates as input.
For project and code/expert/API requests: https://www.catalyzex.com/paper/arxiv:2007.11678.
This process produces motions that are significantly more realistic than those from purely kinematic methods, substantially improving quantitative measures of both kinematic and dynamic plausibility.

Posted by uniqueone
,

Latest from Max Planck researchers: State of the art in Shape and Pose Disentanglement for 3D Meshes!

For project and code/expert/API requests: https://www.catalyzex.com/paper/arxiv:2007.11341

The experiments on datasets of 3D humans, faces, hands and animals demonstrate the generality of our approach.

Posted by uniqueone
,

From #ECCV2020: Reconstruct a morphable shape, texture, and viewpoint from an image collection without 3D ground truth *and* 2D keypoints, allowing us to explore new categories like shoes!

For project and code/expert/API requests: https://www.catalyzex.com/paper/arxiv:2007.10982

Posted by uniqueone
,

Facebook released a Machine Learning algorithm (Multilevel Pixel Aligned Implicit Function For High Resolution 3D Human Digitization.) that can reconstruct or generate 3D pose by just looking at a single image. So, is this a replacement of Character Designers. What do you think guys?

Paper: https://arxiv.org/pdf/2004.00452.pdf
Code: https://github.com/facebookresearch/pifuhd
Colab: https://colab.research.google.com/drive/11z58bl3meSzo6kFqkahMa35G5jmh2Wgt

Credit: PIFuHD or Multilevel Pixel Aligned Implicit Function For Heigher Resolution 3D Human Digitization.

#MachineLearning #machinelearningalgorithms #facebookpost #AI #artificialintelligence #facebookページ

Posted by uniqueone
,

Latest from Microsoft researchers: High-quality video inpainting!

For project and code/expert/API requests: https://www.catalyzex.com/paper/arxiv:2007.10247

They propose to learn a joint Spatial-Temporal Transformer Network (STTN) for video inpainting. Specifically, they simultaneously fill missing regions in all input frames by self-attention, and propose to optimize STTN by a spatial-temporal adversarial loss

Posted by uniqueone
,

Latest from Baidu researchers: Automatic video inpainting algorithm that can remove traffic agents from videos and synthesize missing regions with the guidance of depth/point cloud!

For project and code/expert/API requests: https://www.catalyzex.com/paper/arxiv:2007.08854

In order to fill a target inpainting area in a frame, it is straightforward to transform pixels from other frames into the current one with correct occlusion.

Posted by uniqueone
,

Agile Human Behavior Imitation by humanoid models!

For project and code/API/expert requests: https://www.catalyzex.com/paper/arxiv:2006.07364

This approach is the first humanoid control method that successfully learns from a large-scale human motion dataset (Human3.6M) and generates diverse long-term motions.

Posted by uniqueone
,

Better than Faceapp: State of the art in Facial Attribute Editing!

For project and code/API/expert requests: https://www.catalyzex.com/paper/arxiv:2007.05892

Existing approaches suffer from a serious compromise between correct attribute generation and preservation of the other information such as identity and background, because they edit the attributes in the imprecise area. To resolve this dilemma, we propose a progressive attention GAN (PA-GAN) for facial attribute editing.

Posted by uniqueone
,

Latest from Adobe and UC Berkeley researchers: State of the art in deep image manipulation.
For project and code/API/dataset requests: https://www.catalyzex.com/paper/arxiv:2007.00653
The key idea is to encode an image into two independent components and enforce that any swapped combination maps to a realistic image.

Posted by uniqueone
,

Deep single image manipulation using conditional adversarial generators!
For project and code/API/dataset requests: https://www.catalyzex.com/paper/arxiv:2007.01289
Their network learns to map between a primitive representation of the image (e.g. edges) to the image itself. At manipulation time, their generator allows for making general image changes by modifying the primitive input representation and mapping it through the network.

Posted by uniqueone
,

오늘 소개드릴 논문은 볼수록 여러 BIZ에 적용할 만한 사업 아이디어가 많이 떠올랐던 재미있는 논문이었습니다.
[동영상1]은 드라마에서 캡쳐한 '김치싸대기'를 100번 반복해놓은 영상( https://www.youtube.com/watch?v=1fdtnml--Nc )이 정말 100번 반복을 잘했는지 확인해본 영상이고 정말 100번이 맞았습니다. 물론 직접 카운트 한 것이 아니라 아래 소개드릴 논문을 기반으로 한 코드가 카운트 해주었습니다.(참고로 영상의 숫자는 원본에 없었고 프로그램으로 넣었습니다.)

Counting Out Time: Class Agnostic Video Repetition Counting in the Wild (CVPR 2020)
심장박동, 스쿼트, 지구 자전, 공장제조라인, 연탄나르기, 교통 등의 공통점은 특정 행동이나 현상이 반복적으로 발생한 다는 것입니다.
이러한 현상이 얼마나 자주 어떻게 발생하는지 자동으로 분석할 수 있으면 여러가지 분야에서(공장, CCTV, 헬스케어 등) 활용할 수 있을 것입니다.
문제는 이런 반복현상(action unit)이 아주 똑같이 반복한다면 template 기반으로 쉽게 구현할 수 있을것 같은데 조명, 불필요한 움직임, 변화하는 속도 등 여러 noise에 의해서 난이도가 매우 높아집니다. 이러한 문제를 해결하기 위해서 DeepMind에서 새로운 논문을 냈습니다.
이 논문은 반복적으로 도구를 사용하는 사람, 날개를 펄럭이는 새, 진자 등의 다양한 분야의 반복적인 행위를 캐치할 수 있는 RepNet를 제시합니다.
RepNet architecture
논문을 보기전에 Periodicity Estimation, Temporal Self-similarity Matrix 두 가지에 대해서는 미리 보시면 좋을 것 같습니다.
RepNet은 첫 번째 그림과 같이 하나의 비디오로부터 이미지를 뽑아내고 각 이미지에 대하여 self-similarity를 구한 후 반복의 주기와 길이를 예측 후 반복 장면을 Count를 하는 논문입니다.
[그림2]을 보시면 각 단계에 대해서 자세히 나와있습니다. 우선 Video 파일을 encoder를 거쳐서 각 프레임을 임베딩합니다. 그 다음 Temporal Self_simliarity Matrix를 구하게 되는데 이 simiarity는 각 장면을 가로와 세로로 배치한 후 유사도를 구합니다. 그럼 같은 프레임끼리는 일치하게되므로 대각선으로 노란선이 명확히 생길 것이고 중요하게 봐야할 부분이 그 옆에 노란+초록색 대각선들입니다. [그림1]에서 처럼 이렇게 반복되는 것의 길이를 구할 수 있고 그 카운트도 셀 수 있습니다. 물론 네모 외곽들이 흐린 이유는 반복의 시작과 마지막 부분이라 겹치는 부분이 적어서 입니다.
여기까지는 이전에도 유사한 시도들은 많았지만 [그림3]의 c-d와 같이 속도가 변화하거나 2가지 동작이 섞여 있을때 자동으로 인지하는 것은 어렵습니다. 그래서 Prediod Predictor 부분을 TSM을 업샘플링하여 2D 라인을 32채널로 만들고 (64*32) 각 단계를 Transformer를 통해 Preriod Length Predictor와 Periodicity Predictor로 보내서 최종 loss를 최적화하는 쪽으로 트레이닝을 시켰습니다.
트레이닝을 위해서 오픈된 데이터셋과 임의의 반복 영상을 넣고 Camera Motion Augment도 사용했습니다.
One Model, Many Domains and applications
예전에 일반적인 비전 모델들과 달리 이 모델은 Repetition Counting, Periodicity Detection, Change Inspection, Speed Change Detection, Cross-Period Retrieval 영역에서 모두 활용 할 수 있습니다. 여기서 좀 더 소설을 써보자면 Repetition Counting와 Speed Change Detection을 통해 집에서 카메라로 AI로 부터 PT를 받거나("푸쉬업 30번중에 15번 남았습니다. 속도가 점점 떨어지니 다음 rep은 20번만 하겠습니다."와 같은 것을 스스로 측정 후 가이드), Change Inspection를 이용해서 특정 공정중에 발생하는 예상치못한 문제 확인 등을 해볼 수 있지 않을까 합니다.
그래서 아래 첨부한 영상외에 여러가지를 테스트 해보았는데 추후 이 팀에서 나오는 논문이 매우 기대됩니다. 참고로 작년에 나온 Temporal Cycle-Consistency Learning(CVPR 2019)도 재미있게 봤었는데 올해는 단일 비디오에서 더 다양한 영역이 가능하다는 점에서 많은 발전이 있었습니다.

Posted by uniqueone
,

A Neural Rendering Framework for Free-Viewpoint Relighting

중국 상하이에서 나온 뉴럴 렌더링 논문입니다.

여러 뷰에서 잡은 이미지로 학습하면 임의의 뷰에서 본 렌더링 영상을 조명상태까지 바꾸면서 생성할 수 있답니다.

PyTorch로 구현한 소스코드도 깃허브에 함께 공개되었네요. 요즘 연구분야의 인공지능 프레임웍은 PyTorch가 완전 대세인 듯합니다.

논문
pdf: [https://arxiv.org/pdf/1911.11530.pdf](https://l.facebook.com/l.php?u=https%3A%2F%2Farxiv.org%2Fpdf%2F1911.11530.pdf%3Ffbclid%3DIwAR204MKI_b3-LADBmc6-sFA648X8s-1DE96uboQQrU-4AB-9FQ52P17LRRA&h=AT2Iky7XUG6KjOxvcofYIJQCGp4bhx1x2OqPbYx5Kre8YQTvEEcYNaEcqqGTIRwiZkACVc26Ds9LpJZASUxSweh8sxWdsfZ-Rpn2R7rH9vsneHd7UZgaPHaNUgMtLAMI5EyfLFpYeQA5KyZigpdS5DyvWmQw2ZOnFH_J)
abs: [https://arxiv.org/abs/1911.11530](https://l.facebook.com/l.php?u=https%3A%2F%2Farxiv.org%2Fabs%2F1911.11530%3Ffbclid%3DIwAR2IbAM_XVMoTd_GHUxoUwx4q__ECjW14lTC0VfA1AbVGL3PAWdp1ZEEmWk&h=AT1Unl6kuXEO2NGD7DG7pBJS_B-vfjYTVGXIk3xlbtWBFLeSJY5aYf1N4S9tjoNIOTntmWl4ZZCRYKxqW7ZKcIoXzBobCRuweSP6hh5NYAqrrazoPUp-6uBSN2AGuJWW49I05Gman77rR5dOd8OMUxxOZ086Jfm8MXo8)

PyTorch 소스코드
github: [https://github.com/LansburyCH/relightable-nr](https://l.facebook.com/l.php?u=https%3A%2F%2Fgithub.com%2FLansburyCH%2Frelightable-nr%3Ffbclid%3DIwAR3N9jfs1VAEnO75j4P7mhT0xCLF2m5Mag4TwU2jWAJPfcsU6dj4g23Um90&h=AT0dlocS6x5AzhZUs9jZ9Rk19ojKxI9WnTLJrOEETBcieZurG04t4b40wgLZhLpPrQaQSPqK96Zdn6ILlpuTw30tkjvFMt58YThbTUirytLUkjXUpLK2Pbex7SvCwDiyjioazp5TjtvRT0-UvQGRUCFjh5toM_SC-Hmw)

Posted by uniqueone
,

Faces à la Carte: Text-to-Face Generation via Attribute Disentanglement

얼굴을 묘사한 텍스트를 입력받아 얼굴 이미지를 생성하는 신경망입니다.

이름하여 TTF(Text To Face)라는 흥미로운 연구네요.

pdf: https://arxiv.org/pdf/2006.07606.pdf
abs: https://arxiv.org/abs/2006.07606

Posted by uniqueone
,

안녕하세요 FAIR에서 일하고 있는 주한별입니다. 이번에 CVPR oral에 발표될 PIFuHD라는 논문을 소개해드립니다. 이미지 한장으로 사람의 3D를 복원하는 연구인데, 제가 진행해봤던 연구중에 결과가 가장 신기(?)했던 연구라서 shamelessly 공유합니다. 뒷모습도 복원이 됩니다. 공유하는 동영상은 각 frame을 따로 processing한 결과입니다.

Paper: https://arxiv.org/pdf/2004.00452.pdf

Code: https://github.com/facebookresearch/pifuhd

Project page: https://shunsukesaito.github.io/PIFuHD/

Colab demo: https://colab.research.google.com/drive/11z58bl3meSzo6kFqkahMa35G5jmh2Wgt

Colab demo를 통해 직접 사진을 업로드 하고 3D복원 결과를 볼수 있습니다. 혹시 재미있는 결과를 얻으셨다면 FB이나 Twitter로 공유해주시면 (#pifuhd) 다음 연구진행할때 적극 참고하겠습니다...!

Posted by uniqueone
,

Latest from Microsoft researchers: Recovering the 3D geometry of human head from a single portrait image

For project and code/API/expert requests: https://www.catalyzex.com/paper/arxiv:2004.11598

method is learned in an unsupervised manner without any ground-truth 3D data.

Posted by uniqueone
,

Interesting Research!!!
S2IGAN — Speech-to-Image Generation via Adversarial Learning

Authors present a framework that translates speech to images bypassing text information, thus allowing unwritten languages to potentially benefit from this technology.

ArXiV: https://arxiv.org/abs/2005.06968
Project: https://xinshengwang.github.io/project/s2igan/

Posted by uniqueone
,

From SIGGRAPH 2020: Method reconstructs the geometry of complex 3D thin structures in high quality from a color video captured by a handheld camera

For project and code/API/expert request: [https://www.catalyzex.com/paper/arxiv:2005.03372](https://www.catalyzex.com/paper/arxiv:2005.03372)

Method achieves accurate camera pose estimation and faithful reconstruction of 3D thin structures with complex shape and topology at a level that has not been attained by other existing reconstruction methods.

Posted by uniqueone
,

Reconstruct 3D human body shapes based on a sparse set of RGBD frames using a single RGBD camera

For project or code/API/expert requests: https://www.catalyzex.com/paper/arxiv:2006.03630

Empirical evaluations on synthetic and real datasets demonstrate both quantitatively and qualitatively the superior performance of our framework in reconstructing complete 3D human models with high fidelity.

Posted by uniqueone
,

Recapture your portrait photos with desired posture/view, figure, and clothing style!

For project and API or code request: https://www.catalyzex.com/paper/arxiv:2006.01435

It can properly infer invisible body parts and clothes in original portraits, e.g. the lower body, and meanwhile guarantee global coherency of different regions in recaptured portraits.

Posted by uniqueone
,

Latest from Samsung researchers: State of the art in photo editing (Harmonization)

For project and code or API request: https://www.catalyzex.com/paper/arxiv:2006.00809

They create the models as a combination of existing encoder-decoder architectures and a pre-trained foreground-aware deep high-resolution network.

Posted by uniqueone
,

DeepFaceDrawing: Deep Generation of Face Images from Sketches
Paper: http://geometrylearning.com/paper/DeepFaceDrawing.pdf
Video: https://www.youtube.com/watch?v=HSunooUTwKs

Posted by uniqueone
,

היי חברים,
אני מעביר הרצאות בנושא של מידול תלת מימדי מתמונה בעזרת רשתות ניורנים. בשבועות האחרונים התחלתי להעביר את זה בזום והייתה היענות גבוה (בעיקר מהקהילה החדשה שנוצרה לנו ברדיט https://www.reddit.com/r/2D3DAI/ ), לכן החלטתי לפתוח עוד 4 הרצאות השבוע, המעוניינים מוזמנים להצטרף:

(technical) - June 5th 09:30
https://www.reddit.com/r/2D3DAI/comments/gu06j6/from_2d_to_3d_using_artificial_intelligence_east/

(technical) - June 5th 20:30
https://www.reddit.com/r/2D3DAI/comments/gu097x/from_2d_to_3d_using_artificial_intelligence_west/

(semi-technical) - June 4th 09:30
https://www.reddit.com/r/2D3DAI/comments/gu072a/from_2d_to_3d_using_artificial_intelligence_east/

(semi-technical) - June 4th 20:30
https://www.reddit.com/r/2D3DAI/comments/gu07uq/from_2d_to_3d_using_artificial_intelligence_west/

הרצאות technical דורשות היכרות בסיסית עם רשתות ניורנוים CNN, ResNet. הsemi-technical זה תוכן דומה עם הקדמה על מה זה CNN וResNet וקצת פחות צלילה לפרטים.
כל ההרצאות יהיו באנגלית.

Posted by uniqueone
,

Latest from Apple researchers: Deep learning approach for driving animated faces using both acoustic and visual information

For project and code or API requests: https://www.catalyzex.com/paper/arxiv:2005.13616

To ensure that the model exploits both modalities during training, batches are generated that contain audio-only, video-only, and audiovisual input features

Posted by uniqueone
,

From Adobe researchers: State of the art in High-Resolution Image Inpainting For project and code or API request: https://www.catalyzex.com/paper/arxiv:2005.11742

To mimic real object removal scenarios, they collect a large object mask dataset and synthesize more realistic training data that better simulates user inputs

Posted by uniqueone
,

https://www.facebook.com/deeplearning101/posts/3637994496216575

 

Adversarial Colorization of Icons Based... - Deep Learning London | Facebook

Adversarial Colorization of Icons Based on Structure and Color Conditions Authors: Tsai-Ho Sun, Chien-Hsun Lai, Sai-Keung Wong, and Yu-Shuen Wang Abstract: We present a system to help #designers create icons that are widely used in banners, signboards, bil

www.facebook.com

 

Adversarial Colorization of Icons Based on Structure and Color Conditions

Authors: Tsai-Ho Sun, Chien-Hsun Lai, Sai-Keung Wong, and Yu-Shuen Wang

Abstract: We present a system to help #designers create icons that are widely used in banners, signboards, billboards, homepages, and #mobile apps. Designers are tasked with drawing contours, whereas our system colorizes contours in different styles. This goal is achieved by training a dual conditional generative adversarial network (GAN) on our collected icon dataset.

Source:

Pdf: https://t.co/6tIoJZiXye

Abs: https://t.co/2LakM2d1bk

Github: https://t.co/hV7v3wlzvU

Posted by uniqueone
,