Presentation | 2022-12-15 Training Method for Image-based Instance Segmentation by Video-based Object-Centric Representation Learning Tomokazu Kaneko, Ryosuke Sakai, Soma Shiraishi, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Object-centric representation learning (OCRL) aims to separate and extract object-wise representations from an image. There are two types of methods: image-based and video-based. The former is applicable to a static scene/image but hardly handles complex textures, while the latter is only applicable to videos yet handles textures better using object motions to separate them. We propose a novel image-based approach capable of handling complex textures in crowded scenes as effectively as video-based methods. Our approach generates pseudo labels by a video-based model on which we train an image-based model. We show its effectiveness on instance segmentation task by experiments. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Object-Centric Representation Learning / Instance Segmantation / Unsupervised Learning |
Paper # | PRMU2022-40 |
Date of Issue | 2022-12-08 (PRMU) |
Conference Information | |
Committee | PRMU |
---|---|
Conference Date | 2022/12/15(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Toyama International Conference Center |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Seiichi Uchida(Kyushu Univ.) |
Vice Chair | Takuya Funatomi(NAIST) / Mitsuru Anpai(Denso IT Lab.) |
Secretary | Takuya Funatomi(CyberAgent) / Mitsuru Anpai(Univ. of Tokyo) |
Assistant | Nakamasa Inoue(Tokyo Inst. of Tech.) / Yasutomo Kawanishi(Riken) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Training Method for Image-based Instance Segmentation by Video-based Object-Centric Representation Learning |
Sub Title (in English) | |
Keyword(1) | Object-Centric Representation Learning |
Keyword(2) | Instance Segmantation |
Keyword(3) | Unsupervised Learning |
1st Author's Name | Tomokazu Kaneko |
1st Author's Affiliation | NEC Corporation(NEC) |
2nd Author's Name | Ryosuke Sakai |
2nd Author's Affiliation | NEC Corporation(NEC) |
3rd Author's Name | Soma Shiraishi |
3rd Author's Affiliation | NEC Corporation(NEC) |
Date | 2022-12-15 |
Paper # | PRMU2022-40 |
Volume (vol) | vol.122 |
Number (no) | PRMU-314 |
Page | pp.pp.43-48(PRMU), |
#Pages | 6 |
Date of Issue | 2022-12-08 (PRMU) |