Multi categorical of common eye disease detect using convolutional neural network: a transfer learning approach

Abu Kowshir Bitto, Imran Mahmud

Abstract


Among the most important systems in the body is the eyes. Although their small stature, humans are unable to imagine existence without it. The human optic is safe against dust particles by a narrow layer called the conjunctiva. It prevents friction during the opening and shutting of the eye by acting as a lubricant. A cataract is an opacification of the eye's lens. There are various forms of eye problems. Because the visual system is the most important of the four sensory organs, external eye abnormalities must be detected early. The classification technique can be used in a variety of situations. A few of these uses are in the healthcare profession. We use visual geometry group (VGG-16), ResNet-50, and Inception-v3 architectures of convolutional neural networks (CNNs) to distinguish between normal eyes, conjunctivitis eyes, and cataract eyes throughout this paper. With a detection time of 485 seconds, Inception-v3 is the most accurate at detecting eye disease, with a 97.08% accuracy, ResNet-50 performs the second-highest accuracy with 95.68% with 1090 seconds and lastly, VGG-16 performs 95.48% accuracy taking the highest time of 2510 seconds to detect eye diseases.

Keywords


Eye disease; Inception v3; ResNet50; Transfer learning; VGG16

Full Text:

PDF


DOI: https://doi.org/10.11591/eei.v11i4.3834

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Bulletin of EEI Stats