Autonomous novel class discovery for vision-based recognition in non-interactive environments

Visual recognition with deep learning has recently been shown to be effective in robotic vision. However, these algorithms tend to be build under fixed and structured environment, which is rarely the case in real life. When facing unknown objects, avoidance or human interactions are required, which...

Full description

Saved in:
Bibliographic Details
Main Authors: Xuelin Zhang, Feng Liu, Xuelian Cheng, Siyuan Yan, Zhibin Liao, Zongyuan Ge
Format: Article
Language:English
Published: KeAi Communications Co. Ltd. 2024-01-01
Series:Cognitive Robotics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2667241324000156
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846122016051036160
author Xuelin Zhang
Feng Liu
Xuelian Cheng
Siyuan Yan
Zhibin Liao
Zongyuan Ge
author_facet Xuelin Zhang
Feng Liu
Xuelian Cheng
Siyuan Yan
Zhibin Liao
Zongyuan Ge
author_sort Xuelin Zhang
collection DOAJ
description Visual recognition with deep learning has recently been shown to be effective in robotic vision. However, these algorithms tend to be build under fixed and structured environment, which is rarely the case in real life. When facing unknown objects, avoidance or human interactions are required, which may miss critical objects or be prohibitively costly to obtain on robots in the real world. We consider a practical problem setting that aims to allow robots to automatically discover novel classes with only labelled known class samples in hand, defined as open-set clustering (OSC). To address the OSC problem, we propose a framework combining three approaches: 1) using selfsupervised vision transformers to mitigate the discard of information needed for clustering unknown classes; 2) adaptive weighting for image patches to prioritize patches with richer textures; and 3) incorporating a temperature scaling strategy to generate more separable feature embeddings for clustering. We demonstrate the efficacy of our approach in six fine-grained image datasets.
format Article
id doaj-art-30de12f518704b5e8becfb2a6b3b6813
institution Kabale University
issn 2667-2413
language English
publishDate 2024-01-01
publisher KeAi Communications Co. Ltd.
record_format Article
series Cognitive Robotics
spelling doaj-art-30de12f518704b5e8becfb2a6b3b68132024-12-15T06:17:54ZengKeAi Communications Co. Ltd.Cognitive Robotics2667-24132024-01-014191203Autonomous novel class discovery for vision-based recognition in non-interactive environmentsXuelin Zhang0Feng Liu1Xuelian Cheng2Siyuan Yan3Zhibin Liao4Zongyuan Ge5Faculty of Information Technology, Monash University, 20 Exhibition Walk, Melbourne 3168, VIC, Australia; Corresponding author.The School of Computing and Information Systems, The University of Melbourne, Parkville, Melbourne 3052, VIC, AustraliaFaculty of Information Technology, Monash University, 20 Exhibition Walk, Melbourne 3168, VIC, AustraliaFaculty of Information Technology, Monash University, 20 Exhibition Walk, Melbourne 3168, VIC, AustraliaFaculty of Sciences, Engineering and Technology, The University of Adelaide, North Terrace, Adelaide, 5001, SA, AustraliaFaculty of Information Technology, Monash University, 20 Exhibition Walk, Melbourne 3168, VIC, AustraliaVisual recognition with deep learning has recently been shown to be effective in robotic vision. However, these algorithms tend to be build under fixed and structured environment, which is rarely the case in real life. When facing unknown objects, avoidance or human interactions are required, which may miss critical objects or be prohibitively costly to obtain on robots in the real world. We consider a practical problem setting that aims to allow robots to automatically discover novel classes with only labelled known class samples in hand, defined as open-set clustering (OSC). To address the OSC problem, we propose a framework combining three approaches: 1) using selfsupervised vision transformers to mitigate the discard of information needed for clustering unknown classes; 2) adaptive weighting for image patches to prioritize patches with richer textures; and 3) incorporating a temperature scaling strategy to generate more separable feature embeddings for clustering. We demonstrate the efficacy of our approach in six fine-grained image datasets.http://www.sciencedirect.com/science/article/pii/S2667241324000156Open setImage recognitionImage clusteringDeep learningDeep neural networks
spellingShingle Xuelin Zhang
Feng Liu
Xuelian Cheng
Siyuan Yan
Zhibin Liao
Zongyuan Ge
Autonomous novel class discovery for vision-based recognition in non-interactive environments
Cognitive Robotics
Open set
Image recognition
Image clustering
Deep learning
Deep neural networks
title Autonomous novel class discovery for vision-based recognition in non-interactive environments
title_full Autonomous novel class discovery for vision-based recognition in non-interactive environments
title_fullStr Autonomous novel class discovery for vision-based recognition in non-interactive environments
title_full_unstemmed Autonomous novel class discovery for vision-based recognition in non-interactive environments
title_short Autonomous novel class discovery for vision-based recognition in non-interactive environments
title_sort autonomous novel class discovery for vision based recognition in non interactive environments
topic Open set
Image recognition
Image clustering
Deep learning
Deep neural networks
url http://www.sciencedirect.com/science/article/pii/S2667241324000156
work_keys_str_mv AT xuelinzhang autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments
AT fengliu autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments
AT xueliancheng autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments
AT siyuanyan autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments
AT zhibinliao autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments
AT zongyuange autonomousnovelclassdiscoveryforvisionbasedrecognitioninnoninteractiveenvironments