Discriminative kernel convolution network for multi-label ophthalmic disease detection on imbalanced fundus image dataset

Amit Bhati, Neha Gour, Pritee Khanna, Aparajita Ojha

Abstract: Highlights•Multi-label ophthalmic disease detection on an imbalanced fundus image dataset.•Discriminative kernel convolution network for ophthalmic disease classification.•Dilated convolution-based attention network to detect multiple lesions simultaneously.•A class balancing factor to dynamically resolve the class imbalance issue.•The method outperforms eight state-of-the-art methods with fewer computations.