TY - GEN
T1 - CNN-Based Obstacle Avoidance Using RGB-Depth Image Fusion
AU - Mechal, Chaymae El
AU - El Idrissi, Najiba El Amrani
AU - Mesbah, Mostefa
N1 - Funding Information:
The authors would thank the Department of Electrical and Computer Engineering of Sultan Qaboos University for hosting the first author during the work on the project that let to this paper.
Publisher Copyright:
© 2022, Springer Nature Singapore Pte Ltd.
PY - 2022
Y1 - 2022
N2 - In the last few years, deep learning has attracted wide interest and achieved great success in many computer vision related applications, such as image classification, object detection, object tracking, pose estimation and action recognition. One specific application that can greatly benefit from the recent advance of deep learning is robot vision-based obstacle avoidance. Vision-based obstacle avoidance systems are mostly based on classification algorithms. Most of these algorithms use either color images or depth images as the main source of information. In this paper, the aim is to investigate whether using information extracted from both types of images simultaneously would give better performance than using each one separately. To do this, we chose the convolutional neural network (CNN) as the classifier and HSV-based method to achieve the fusion. We tested this approach using two widely used pre-trained CNN architectures, namely Resnet-50 and GoogLeNet using a dataset locally collected. The results indicate that the image fusion-based classification algorithm achieve a higher accuracy (91.3%) than the one based on depth images (80.4%) but lower than the one based on color images (93.7%). These results can be partly explained by the fact that the used classifiers were pre-trained using color image datasets.
AB - In the last few years, deep learning has attracted wide interest and achieved great success in many computer vision related applications, such as image classification, object detection, object tracking, pose estimation and action recognition. One specific application that can greatly benefit from the recent advance of deep learning is robot vision-based obstacle avoidance. Vision-based obstacle avoidance systems are mostly based on classification algorithms. Most of these algorithms use either color images or depth images as the main source of information. In this paper, the aim is to investigate whether using information extracted from both types of images simultaneously would give better performance than using each one separately. To do this, we chose the convolutional neural network (CNN) as the classifier and HSV-based method to achieve the fusion. We tested this approach using two widely used pre-trained CNN architectures, namely Resnet-50 and GoogLeNet using a dataset locally collected. The results indicate that the image fusion-based classification algorithm achieve a higher accuracy (91.3%) than the one based on depth images (80.4%) but lower than the one based on color images (93.7%). These results can be partly explained by the fact that the used classifiers were pre-trained using color image datasets.
KW - Convolutional neural network
KW - Deep learning
KW - Image fusion
KW - Obstacle avoidance
KW - Robot vision
UR - http://www.scopus.com/inward/record.url?scp=85113303068&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85113303068&partnerID=8YFLogxK
U2 - 10.1007/978-981-33-6893-4_78
DO - 10.1007/978-981-33-6893-4_78
M3 - Conference contribution
AN - SCOPUS:85113303068
SN - 9789813368927
T3 - Lecture Notes in Electrical Engineering
SP - 867
EP - 876
BT - WITS 2020 - Proceedings of the 6th International Conference on Wireless Technologies, Embedded, and Intelligent Systems
A2 - Bennani, Saad
A2 - Lakhrissi, Younes
A2 - Khaissidi, Ghizlane
A2 - Mansouri, Anass
A2 - Khamlichi, Youness
PB - Springer Science and Business Media Deutschland GmbH
T2 - 6th International Conference on Wireless Technologies, Embedded and Intelligent Systems, WITS 2020
Y2 - 14 October 2020 through 16 October 2020
ER -