RORD: A Real-world Object Removal Dataset

Min Cheol Sagong; Yoon Jae Yeo; Seung Won Jung; Sung Jea Ko

RORD: A Real-world Object Removal Dataset

Min Cheol Sagong, Yoon Jae Yeo, Seung Won Jung, Sung Jea Ko

Research output: Contribution to conference › Paper › peer-review

Abstract

Various convolutional neural networks (CNNs)-based image inpainting techniques have been actively studied to remove unwanted objects or restore missing parts in recent years. The common standard for training image inpainting CNNs is synthesising hole regions on the existing datasets, such as ImageNet and Places2. However, from the viewpoint of the object removal task, such a methodology is suboptimal because actual pixels behind objects, i.e., “ground truth”, cannot be used for training. Facing this problem, we introduce Real-world Object Removal Dataset (RORD), a large-scale collection of image pairs with and without objects. RORD consists of a wide range of real-world scenes, plus two types of pixel-accurate annotations, i.e., object mask and segmentation map. Our dataset allows existing image inpainting models to be trained accurately as well as evaluated with high confidence. In this paper, we describe in detail how the dataset is constructed and demonstrate the validity and usability of RORD.

Original language	English
Publication status	Published - 2022
Event	33rd British Machine Vision Conference Proceedings, BMVC 2022 - London, United Kingdom Duration: 2022 Nov 21 → 2022 Nov 24

Conference

Conference	33rd British Machine Vision Conference Proceedings, BMVC 2022
Country/Territory	United Kingdom
City	London
Period	22/11/21 → 22/11/24

Bibliographical note

Publisher Copyright:
© 2022. The copyright of this document resides with its authors. It may be distributed unchanged freely in print or electronic forms.

ASJC Scopus subject areas

Computer Vision and Pattern Recognition

Cite this

@conference{a02ec81442714077bea3b9b9c37f0387,

title = "RORD: A Real-world Object Removal Dataset",

abstract = "Various convolutional neural networks (CNNs)-based image inpainting techniques have been actively studied to remove unwanted objects or restore missing parts in recent years. The common standard for training image inpainting CNNs is synthesising hole regions on the existing datasets, such as ImageNet and Places2. However, from the viewpoint of the object removal task, such a methodology is suboptimal because actual pixels behind objects, i.e., “ground truth”, cannot be used for training. Facing this problem, we introduce Real-world Object Removal Dataset (RORD), a large-scale collection of image pairs with and without objects. RORD consists of a wide range of real-world scenes, plus two types of pixel-accurate annotations, i.e., object mask and segmentation map. Our dataset allows existing image inpainting models to be trained accurately as well as evaluated with high confidence. In this paper, we describe in detail how the dataset is constructed and demonstrate the validity and usability of RORD.",

author = "Sagong, {Min Cheol} and Yeo, {Yoon Jae} and Jung, {Seung Won} and Ko, {Sung Jea}",

note = "Publisher Copyright: {\textcopyright} 2022. The copyright of this document resides with its authors. It may be distributed unchanged freely in print or electronic forms.; 33rd British Machine Vision Conference Proceedings, BMVC 2022 ; Conference date: 21-11-2022 Through 24-11-2022",

year = "2022",

language = "English",

}

TY - CONF

T1 - RORD

T2 - 33rd British Machine Vision Conference Proceedings, BMVC 2022

AU - Sagong, Min Cheol

AU - Yeo, Yoon Jae

AU - Jung, Seung Won

AU - Ko, Sung Jea

PY - 2022

Y1 - 2022

N2 - Various convolutional neural networks (CNNs)-based image inpainting techniques have been actively studied to remove unwanted objects or restore missing parts in recent years. The common standard for training image inpainting CNNs is synthesising hole regions on the existing datasets, such as ImageNet and Places2. However, from the viewpoint of the object removal task, such a methodology is suboptimal because actual pixels behind objects, i.e., “ground truth”, cannot be used for training. Facing this problem, we introduce Real-world Object Removal Dataset (RORD), a large-scale collection of image pairs with and without objects. RORD consists of a wide range of real-world scenes, plus two types of pixel-accurate annotations, i.e., object mask and segmentation map. Our dataset allows existing image inpainting models to be trained accurately as well as evaluated with high confidence. In this paper, we describe in detail how the dataset is constructed and demonstrate the validity and usability of RORD.

AB - Various convolutional neural networks (CNNs)-based image inpainting techniques have been actively studied to remove unwanted objects or restore missing parts in recent years. The common standard for training image inpainting CNNs is synthesising hole regions on the existing datasets, such as ImageNet and Places2. However, from the viewpoint of the object removal task, such a methodology is suboptimal because actual pixels behind objects, i.e., “ground truth”, cannot be used for training. Facing this problem, we introduce Real-world Object Removal Dataset (RORD), a large-scale collection of image pairs with and without objects. RORD consists of a wide range of real-world scenes, plus two types of pixel-accurate annotations, i.e., object mask and segmentation map. Our dataset allows existing image inpainting models to be trained accurately as well as evaluated with high confidence. In this paper, we describe in detail how the dataset is constructed and demonstrate the validity and usability of RORD.

UR - http://www.scopus.com/inward/record.url?scp=85174685009&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85174685009

Y2 - 21 November 2022 through 24 November 2022

ER -

RORD: A Real-world Object Removal Dataset

Abstract

Conference

Bibliographical note

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this