vioft2nntf2t|tblJournal|Abstract_paper|0xf4ffa6612c0000004c17060001000500 Deep-Learning and image processing have shown excellent performance in automated fish image classification and recognition task in recent years. In this research paper, we have come up with a novel deep-learning method based on CNN features extracted from deeper layer of a pretrained CNN architecture for automatic classification of eleven (11) indigenous fresh water fish species from India. We have utilized top three layers of a pretrained Resnet-50 model to extract features from fish images and an “ones for all SVM” classifier to train and test images based on the CNN features. This paper reports an exceptional result in overall classification performance on Fish-Pak dataset and on our own dataset. The proposed framework yields overall classification accuracy, precision and recall of 100% on our own data and a maximum of 98.74% accuracy on Fish-Pak dataset which is the best till date.
Jayashree Deka1, Shakuntala Laskar2, Bikramaditya Baklial3
Assam Don Bosco University, India1,Bahona College, India2,3
Automatic Fish Detection, Fish Classification, Fish Species Recognition, Fish Database, Feature Extraction
| January | February | March | April | May | June | July | August | September | October | November | December |
| 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Published By :
ICTACT
Published In :
ICTACT Journal on Image and Video Processing
( Volume: 12 , Issue: 4 , Pages: 2721-2729 )
Date of Publication :
May 2022
Page Views :
105
Full Text Views :
2
|