VNCESM: VGG-Net Based Conjunctive Eyes Segmentation Model with Transfer Learning

The future of healthcare is being altered by technology; non-invasive medical techniques assisted by technology are preferred for medical diagnostics. One of the common illnesses harming people's health worldwide is conjunctivitis, and using cutting-edge technology to treat this condition would greatly lower its prevalence. The goal of this work is to use deep learning to do conjunctiva region segmentation for non-invasive conjunctive eye detection applications. The suggested VGG-net Based Conjunctive Eye Segmentation Model (VNCESM) effectively segments conjunctiva from digital eye pictures taken in an uncontrolled environment by using fine-tuned CNN architecture. Pre-processing and image augmentation were done to boost the model's performance and data volume. UNBCSM demonstrated strong segmentation performance, with segmented masks for training and validation of 97.47%, and 95.49% respectively.