Instance segmentation is a fundamental vision task that aims to recognize and
segment each object in an image. However, it requires costly annotations such
as bounding boxes and segmentation masks for learning. In this work, we propose
a fully unsupervised learning method that learns class-agnostic instance
segmentation without any annotations. We present FreeSOLO, a self-supervised
instance segmentation framework built on top of the simple instance
segmentation method SOLO. Our method also presents a novel localization-aware
pre-training framework, where objects can be discovered from complicated scenes
in an unsupervised manner. FreeSOLO achieves 9.8% AP_50 on the challenging
COCO dataset, which even outperforms several segmentation proposal methods that
use manual annotations. For the first time, we demonstrate unsupervised
class-agnostic instance segmentation successfully. FreeSOLO's box localization
significantly outperforms state-of-the-art unsupervised object
detection/discovery methods, with about 100% relative improvements in COCO AP.
FreeSOLO further demonstrates superiority as a strong pre-training method,
outperforming state-of-the-art self-supervised pre-training methods by +9.8% AP
when fine-tuning instance segmentation with only 5% COCO masks. Code is
available at: github.com/NVlabs/FreeSOLO
Description
FreeSOLO: Learning to Segment Objects without Annotations
%0 Generic
%1 wang2022freesolo
%A Wang, Xinlong
%A Yu, Zhiding
%A De Mello, Shalini
%A Kautz, Jan
%A Anandkumar, Anima
%A Shen, Chunhua
%A Alvarez, Jose M.
%D 2022
%K segmentation
%T FreeSOLO: Learning to Segment Objects without Annotations
%U http://arxiv.org/abs/2202.12181
%X Instance segmentation is a fundamental vision task that aims to recognize and
segment each object in an image. However, it requires costly annotations such
as bounding boxes and segmentation masks for learning. In this work, we propose
a fully unsupervised learning method that learns class-agnostic instance
segmentation without any annotations. We present FreeSOLO, a self-supervised
instance segmentation framework built on top of the simple instance
segmentation method SOLO. Our method also presents a novel localization-aware
pre-training framework, where objects can be discovered from complicated scenes
in an unsupervised manner. FreeSOLO achieves 9.8% AP_50 on the challenging
COCO dataset, which even outperforms several segmentation proposal methods that
use manual annotations. For the first time, we demonstrate unsupervised
class-agnostic instance segmentation successfully. FreeSOLO's box localization
significantly outperforms state-of-the-art unsupervised object
detection/discovery methods, with about 100% relative improvements in COCO AP.
FreeSOLO further demonstrates superiority as a strong pre-training method,
outperforming state-of-the-art self-supervised pre-training methods by +9.8% AP
when fine-tuning instance segmentation with only 5% COCO masks. Code is
available at: github.com/NVlabs/FreeSOLO
@misc{wang2022freesolo,
abstract = {Instance segmentation is a fundamental vision task that aims to recognize and
segment each object in an image. However, it requires costly annotations such
as bounding boxes and segmentation masks for learning. In this work, we propose
a fully unsupervised learning method that learns class-agnostic instance
segmentation without any annotations. We present FreeSOLO, a self-supervised
instance segmentation framework built on top of the simple instance
segmentation method SOLO. Our method also presents a novel localization-aware
pre-training framework, where objects can be discovered from complicated scenes
in an unsupervised manner. FreeSOLO achieves 9.8% AP_{50} on the challenging
COCO dataset, which even outperforms several segmentation proposal methods that
use manual annotations. For the first time, we demonstrate unsupervised
class-agnostic instance segmentation successfully. FreeSOLO's box localization
significantly outperforms state-of-the-art unsupervised object
detection/discovery methods, with about 100% relative improvements in COCO AP.
FreeSOLO further demonstrates superiority as a strong pre-training method,
outperforming state-of-the-art self-supervised pre-training methods by +9.8% AP
when fine-tuning instance segmentation with only 5% COCO masks. Code is
available at: github.com/NVlabs/FreeSOLO},
added-at = {2022-07-17T16:23:49.000+0200},
author = {Wang, Xinlong and Yu, Zhiding and De Mello, Shalini and Kautz, Jan and Anandkumar, Anima and Shen, Chunhua and Alvarez, Jose M.},
biburl = {https://www.bibsonomy.org/bibtex/22321e40b4a995adad230f59a750a85ee/redtedtezza},
description = {FreeSOLO: Learning to Segment Objects without Annotations},
interhash = {5229be1b0fb446cc4d99c2bfafba6286},
intrahash = {2321e40b4a995adad230f59a750a85ee},
keywords = {segmentation},
note = {cite arxiv:2202.12181Comment: 13 pages. Accepted to IEEE/CVF Conf. Comp. Vision Pattern Recognition (CVPR) 2022},
timestamp = {2022-07-17T16:23:49.000+0200},
title = {FreeSOLO: Learning to Segment Objects without Annotations},
url = {http://arxiv.org/abs/2202.12181},
year = 2022
}