🧠📸 OptiPredict: Smart Fundus Imaging and Automated Ocular Disease Diagnosis

OptiPredict is a dual-solution system combining a portable, nonmydriatic fundus camera prototype with a deep learning-powered diagnostic platform. Designed for affordability, mobility, and diagnostic accuracy, this system empowers ophthalmologists and healthcare providers in resource-constrained environments.

---

🔬 Project Highlights

📷 1. Portable Fundus Camera Prototype

A low-cost, handheld, nonmydriatic fundus camera engineered using:

• Raspberry Pi with IR-sensitive camera module
• Dual infrared and white-light LEDs
• 5-inch touchscreen LCD
• Rechargeable battery for portability
• Disposable 20D condensing lens

💡 Specs: 133mm × 91mm × 45mm, 386g, ₹15,374

✅ Key Benefits:

• Captures high-quality retinal images without pharmacologic dilation
• Compact, lightweight, and ideal for rural or mobile screenings
• Usable by healthcare workers with minimal training

---

🤖 2. AI-Powered Ocular Disease Detection (OptiPredict)

A deep learning-based diagnostic platform trained on fundus images to detect eye diseases automatically and in real time.

🧠 Models Implemented:

• ResNet-50 – 98% validation accuracy
• VGG-19 – 94.44% validation accuracy
• Vision Transformer – 85.71% validation accuracy

🧰 Core Technologies:

• Local Binary Patterns (LBP) for enhanced feature extraction
• Binary classification to mitigate class imbalance
• ODIR dataset (6,000 labeled fundus images)
• Cloud-based architecture for deployment & real-time diagnostics

🔁 Preprocessing Pipeline:

• Image normalization, resizing, and augmentation
• Feature enhancement using LBP
• Split into training/validation sets for robust evaluation