Localization of Optic Disc in Fundus Images Based on Yolov3
- Resource Type
- Conference
- Authors
- Jiang, Tianci; Zhou, Zhiwei; Yao, Jun
- Source
- 2021 40th Chinese Control Conference (CCC) Chinese Control Conference (CCC), 2021 40th. :8050-8055 Jul, 2021
- Subject
- Computing and Processing
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Location awareness
Pathology
Optical imaging
Retina
Fundus Image
Optic Disc Localization
Yolov3
- Language
- ISSN
- 1934-1768
In order to improve the detection accuracy and detection speed of optic disc in fundus images, an end-to-end localization method based on Yolov3 framework was proposed in this paper. The pre-processed fundus images were input into Yolov3 framework to locate optic discs accurately. This proposed method achieved 100% accuracy on MESSIDOR, DIARETDB0, DIARETDB1 and DRIVE datasets. 97.5% and 98% success rates were obtained on STARE and KAGGLE datasets respectively. The average localization time was 0.003s. The test results demonstrate that this proposed method can locate optic disc in both normal fundus images and the fundus images with pathological retinas robustly and rapidly.