Scene Classification Based on Light Weight Scene Change Detection for Resource-Constrained Mobile Robots 


Vol. 46,  No. 12, pp. 2450-2457, Dec.  2021
10.7840/kics.2021.46.12.2450


PDF
  Abstract

The mobile robot requires continuous awareness of the context of the spatial environment to carry out given missions precisely. Recently, CNN, a kind of deep learning technology, has been used to classify scenes from camera images to recognize the space. We need lightweight robots so that mobile robots can be used in various applications broadly, but the CNN-based sceneclassifier requires high computation resources and battery consumption. In this paper, we propose a scene classification technique based on scene change detection to reduce computational complexity. The proposed scene classifier shows a similar performance to the previous CNN-based scene classifier, but the computational complexity is about half. The proposed classifier uses a low-complexity scene change detector to reduce complexity. The compute-intensive CNN classifier is only triggered when the scene change is detected. We evaluated the proposed scene classifier using indoor videos to verify whether it can be applied to indoor mobile robots. The analysis results show that the proposed classifier can achieve scene classification with similar accuracy with computing resources of about 45.49%.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

D. Shin and J. Kim, "Scene Classification Based on Light Weight Scene Change Detection for Resource-Constrained Mobile Robots," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 12, pp. 2450-2457, 2021. DOI: 10.7840/kics.2021.46.12.2450.

[ACM Style]

Dong-Hun Shin and Jae-Ho Kim. 2021. Scene Classification Based on Light Weight Scene Change Detection for Resource-Constrained Mobile Robots. The Journal of Korean Institute of Communications and Information Sciences, 46, 12, (2021), 2450-2457. DOI: 10.7840/kics.2021.46.12.2450.

[KICS Style]

Dong-Hun Shin and Jae-Ho Kim, "Scene Classification Based on Light Weight Scene Change Detection for Resource-Constrained Mobile Robots," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 12, pp. 2450-2457, 12. 2021. (https://doi.org/10.7840/kics.2021.46.12.2450)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880