A Study on Dog Breed Identification through Computer Vision and Deep Learning 


Vol. 44,  No. 12, pp. 2323-2331, Dec.  2019
10.7840/kics.2019.44.12.2323


PDF Full-Text
  Abstract

In this paper, we propose a new approach to dog breed identification model through the computer vision and deep-learning and perform the relevant experiments and describe the results. These experiments will be a cornerstone for research to implement ‘smart pet care services’. The identification of dog breed is based on CNN(Convolution Natural Network) model, which shows excellent performance in Feature representation learning. CNN is known to improve performance if unnecessary areas and noise are removed. Therefore, the proposed model combines U-Net to remove all unnecessary areas through Instance Segmentation and performs identification through CNN. The proposed model showed 17.86% higher performance than the CNN alone model, and 91.62% accuracy in identifying eight dog breeds. However, experiments have shown that the CNN model does not clearly identification between two breeds of similar color arrangements or shapes in the image of a dog that shows only a face or body part. These results has another different significance to the Identification accuracy and the direction of our future research to improve it.

  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]

T. Jung and K. Kim, "A Study on Dog Breed Identification through Computer Vision and Deep Learning," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 12, pp. 2323-2331, 2019. DOI: 10.7840/kics.2019.44.12.2323.

[ACM Style]

Tack-hyun Jung and Keecheon Kim. 2019. A Study on Dog Breed Identification through Computer Vision and Deep Learning. The Journal of Korean Institute of Communications and Information Sciences, 44, 12, (2019), 2323-2331. DOI: 10.7840/kics.2019.44.12.2323.

[KICS Style]

Tack-hyun Jung and Keecheon Kim, "A Study on Dog Breed Identification through Computer Vision and Deep Learning," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 12, pp. 2323-2331, 12. 2019. (https://doi.org/10.7840/kics.2019.44.12.2323)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880