3DDEN: Open-Ended 3-D Object Recognition Using Dynamically Expandable Networks

Service robots, in general, have to work independently and adapt to the dynamic changes happening in the environment in real time. One important aspect in such scenarios is to continually learn to recognize newer object categories when they become available. This combines two main research problems,...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on cognitive and developmental systems 2023-06, Vol.15 (2), p.409-418
Main Authors: Jain, Sudhakaran, Kasaei, Hamidreza
Format: Article
Language:English
Subjects:
3-D object recognition
Computational modeling
continual learning (CL)
dynamic network architectures
Feature extraction
Object recognition
open-ended learning
Robots
Task analysis
Three-dimensional displays
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Service robots, in general, have to work independently and adapt to the dynamic changes happening in the environment in real time. One important aspect in such scenarios is to continually learn to recognize newer object categories when they become available. This combines two main research problems, namely, continual learning and 3-D object recognition. Most of the existing research approaches include the use of deep convolutional neural networks (CNNs) focusing on image data sets. A modified approach might be needed for continually learning 3-D object categories. A major concern in using CNNs is the problem of catastrophic forgetting when a model tries to learn a new task. Despite various proposed solutions to mitigate this problem, there still exist some downsides of such solutions, e.g., computational complexity, especially when learning a substantial number of tasks. These downsides can pose major problems in robotic scenarios where real-time response plays an essential role. Toward addressing this challenge, we propose a new deep transfer learning approach based on a dynamic architectural method to make robots capable of open-ended learning about new 3-D object categories. Furthermore, we make sure that the mentioned downsides are minimized to a great extent. Experimental results showed that the proposed model outperformed state-of-the-art approaches with regards to accuracy and also substantially minimizes computational overhead. The code is available online at: https://github.com/sudhakaranjain/3D_DEN .
ISSN:2379-8920
DOI:10.1109/TCDS.2021.3075143