In this project, work has been done to create an application that works with artificial intelligence and deep learning, as this application is based largely on neural network algorithms and reinforcement learning algorithms.
The problem is Eye Diseases
.
Human eye, in humans, specialized sense organ capable of receiving visual images, which are then carried to the brain.
Many eye conditions can cause low vision, blindness, discomfort, and other life-altering conditions. Other conditions may not cause vision loss but result in vision and eye irregularities.
Refractive errors are the most common type of eye condition. Eyeglasses, contact lenses, and LASIK eye surgery can correct them.
how it happens?
Refractive errors occur when your eye's shape doesn't bend light correctly. This keeps the light from focusing on your retina properly. The result is blurred vision.
Glaucoma is a group of eye conditions that damage your optic nerve. These conditions are usually caused by abnormally high pressure in your eye. They are the leading cause of blindness in older people. Treatment options include lowering eye pressure, eye drops, medications, surgery, and laser treatment.
Macular degeneration is an age-related eye condition. It results in loss of the central vision of your eye (macula).
It causes blurred vision or vision loss. It can be treated with minerals and vitamins or surgery.
Keratoconus is a continually developing eye disease that leads to a bulging cornea. Over time, the cornea starts to form into a cone shape. This diverts the light entering your eye and results in distorted vision.
Retinitis is inflammation of the retina, a thin layer of tissue lining the back of the eye. This condition can permanently damage the retina and cause blindness.
Retinal detachment results in peripheral vision loss over time. It can be caused by injuries, retinal inflammation, posterior vitreous detachment, tumors, and eye surgeries.
The dataset consists of Normal
, Diabetic Retinopathy
, Cataract
and Glaucoma
.
retinal images where each class have approximately 1000 images. These images are collected from various sorces like IDRiD, Oculur recognition, HRF etc.
This is the link of dataset on site Kaggle: Here
In this work, one of the most famous and best reinforcement learning algorithms, DensNet121
,
was applied to our data and very good results were obtained.
The Result:
we use on the Project:
- Pythos as Language
- TensorFlow
- keras
- Flask as Framework for deployment.
- HTML
- BootStrap
- Css
In this project, work has been done to create an application that classifies and detects eye disease, as this application works on reinforcement learning based on the convolutional neural network. The application gave very good results.