01
Upload a Retinal Image
Accepted formats: JPG, JPEG, PNG — maximum file size 16 MB
Drag & drop your retinal image here
or click below to browse your files
Accepted formats: JPG, JPEG, PNG — maximum file size 16 MB
or click below to browse your files
An open research initiative making clinical-grade eye disease screening accessible worldwide.
Netra.AI is a revolutionary AI-powered screening system engineered to assist clinicians in the early detection of common but vision-threatening eye diseases. Our deep learning pipeline analyses retinal fundus images with remarkable precision — rivalling specialist ophthalmologist performance on key benchmarks.
Our mission: make eye disease screening accessible, affordable, and accurate — stopping preventable blindness through early intervention, everywhere in the world.
A clouding of the eye's natural lens, causing blurry vision, faded colours, and glare. The leading cause of vision loss globally — yet highly treatable with surgery when caught early.
A diabetes complication that damages retinal blood vessels. Without early detection it progresses to severe vision impairment. Routine screening is critical for all diabetic patients.
Known as the "silent thief of sight," glaucoma damages the optic nerve with no early symptoms, causing irreversible peripheral vision loss. AI-based screening changes the prognosis.
A healthy retina with clear blood vessels, an intact optic disc, and an undamaged macula — the gold-standard baseline against which all conditions are assessed.
From upload to result in under five seconds — fully automated, completely transparent.
Drag and drop or browse for your retinal fundus photograph in JPG or PNG format.
The image is normalised, resized, and enhanced to match the model's training distribution.
Our convolutional neural network classifies the image across four disease categories with confidence scores.
A detailed analysis report — with probability bars, charts, and per-class metrics — is generated instantly.
Everything a clinician or researcher needs for reliable, rapid retinal image analysis.
State-of-the-art CNN trained on thousands of labelled retinal images. Achieves 96.4% accuracy across four disease classes.
Full analysis completes in under five seconds. Interactive confidence charts and probability bars update live.
Per-class precision, recall, and F1-scores are surfaced alongside every result. No black-box outputs.
Images are processed in-session only and never stored permanently. Your patient data stays yours.
The researchers, engineers, and clinicians who built this system.
Full Stack Developer & Model Trainer
Leading the development of Netra.AI and training deep learning models.
Dataset Curator & Model Evaluator
Curating retinal image datasets and evaluating model performance.
Testing Specialist
Ensuring system reliability across various image conditions.
Model Architect & Performance Analyst
Designing neural network architecture and analyzing model performance.