Enhanced Odirnet: Attention-driven And Explainable Deep Neural Network For Robust Diabetic Retinopathy Detection

• Author(s):

  Prof. N. K. Patil | Rushikesh Patil | Sohel Khan | Chaitanya Gangarde | Shivam Kale

• Keywords:

  Diabetic Retinopathy, Deep Learning, CNN, ODIRNet, Fundus Imaging, Medical Image Classification

• Abstract:

  Diabetic Retinopathy (DR) Is A Significant Global Health Concern, Especially In Areas With High Rates Of Diabetes. Early Detection Via Automated Systems Can Reduce The Risk Of Blindness. This Paper Introduces ODIRNet, A Compact Deep Convolutional Neural Network That Efficiently Classifies Retinal Fundus Images. ODIRNet Is Developed From The Ground Up, Incorporating Advanced Feature Extraction Techniques Such As Blue-channel Emphasis And Attention Modules. The Model Was Trained And Tested On The Ocular Disease Intelligent Recognition (ODIR) Dataset, Which Contains 6392 Images. Results Show That ODIRNet Achieves An Accuracy Of 89.70%, Surpassing Models Like VGG16, ResNet50, And MobileNet. Furthermore, The Model Is Integrated Into A Web-based Platform For Real-time Diagnostic Screening, Making It A Practical And Accessible Solution For Clinics With Limited Resources.

---

Other Details

• Paper id:

  IJSARTV12I4105186

• Published in:

  Volume: 12 Issue: 4 April 2026

• Publication Date:

  2026-04-28

Download Article