Photorealistic Image Synthesis for Object Instance Detection

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F19%3A00328593" target="_blank" >RIV/68407700:21230/19:00328593 - isvavai.cz</a>

Výsledek na webu

<a href="https://arxiv.org/abs/1902.03334" target="_blank" >https://arxiv.org/abs/1902.03334</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/ICIP.2019.8803821" target="_blank" >10.1109/ICIP.2019.8803821</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Photorealistic Image Synthesis for Object Instance Detection

Popis výsledku v původním jazyce

We present an approach to synthesize highly photorealistic images of 3D object models, which we use to train a convolutional neural network for detecting the objects in real images. The proposed approach has three key ingredients: (1) 3D object models are rendered in 3D models of complete scenes with realistic materials and lighting, (2) plausible geometric configuration of objects and cameras in a scene is generated using physics simulations, and (3) high photorealism of the synthesized images achieved by physically based rendering. When trained on images synthesized by the proposed approach, the Faster R-CNN object detector achieves a 24% absolute improvement of mAP@.75IoU on Rutgers APC and 11% on LineMod-Occluded datasets, compared to a baseline where the training images are synthesized by rendering object models on top of random photographs. This work is a step towards being able to effectively train object detectors without capturing or annotating any real images. A dataset of 600K synthetic images with ground truth annotations for various computer vision tasks will be released on the project website: thodan.github.io/objectsynth.

Název v anglickém jazyce

Photorealistic Image Synthesis for Object Instance Detection

Popis výsledku anglicky

We present an approach to synthesize highly photorealistic images of 3D object models, which we use to train a convolutional neural network for detecting the objects in real images. The proposed approach has three key ingredients: (1) 3D object models are rendered in 3D models of complete scenes with realistic materials and lighting, (2) plausible geometric configuration of objects and cameras in a scene is generated using physics simulations, and (3) high photorealism of the synthesized images achieved by physically based rendering. When trained on images synthesized by the proposed approach, the Faster R-CNN object detector achieves a 24% absolute improvement of mAP@.75IoU on Rutgers APC and 11% on LineMod-Occluded datasets, compared to a baseline where the training images are synthesized by rendering object models on top of random photographs. This work is a step towards being able to effectively train object detectors without capturing or annotating any real images. A dataset of 600K synthetic images with ground truth annotations for various computer vision tasks will be released on the project website: thodan.github.io/objectsynth.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

Výsledek vznikl pri realizaci vícero projektů. Více informací v záložce Projekty.

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2019

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

2019 IEEE International Conference on Image Processing (ICIP)

ISBN

978-1-5386-6249-6

ISSN

1522-4880

e-ISSN

2381-8549

Počet stran výsledku

5

Strana od-do

66-70

Název nakladatele

IEEE

Místo vydání

Piscataway, NJ

Místo konání akce

Taipei

Datum konání akce

22. 9. 2019

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—