Region Focus Network for Joint Optic Disc and Cup Segmentation

Glaucoma is one of the three leading causes of blindness in the world and is predicted to affect around 80 million people by 2020. The optic cup (OC) to optic disc (OD) ratio (CDR) in fundus images plays a pivotal role in the screening and diagnosis of glaucoma. Existing methods usually crop the optic disc region first, and subsequently perform segmentation in this region. However, these approaches come up with high complexities due to the separate operations. To remedy this issue, we propose a Region Focus Network (RF-Net) that innovatively integrates detection and multi-class segmentation into a unified architecture for end-to-end joint optic disc and cup segmentation with global optimization. The key idea of our method is designing a novel multi-class mask branch which generates a high-quality segmentation in the detected region for both disc and cup. To bridge the connection between the backbone and multi-class mask branch, a Fusion Feature Pooling (FFP) structure is presented to extract features from each level of the pyramid network and fuse them into a final feature representation for segmentation. Extensive experimental results on the REFUGE-2018 challenge dataset and the Drishti-GS dataset show that the proposed method achieves the best performance, compared with competitive approaches reported in the literature and the official leaderboard. Our code will be released soon.

Arbitrary-Oriented Inshore Ship Detection based on Multi-Scale Feature Fusion and Contextual Pooling on Rotation Region Proposals

Inshore ship detection plays an important role in many civilian and military applications. The complex land environment and the diversity of target sizes and distributions make it still challenging for us to obtain accurate detection results. In order to achieve precise localization and suppress false alarms, in this paper, we propose a framework which integrates a multi-scale feature fusion network, rotation region proposal network and contextual pooling together. Specifically, in order to describe ships of various sizes, different convolutional layers are fused to obtain multi-scale features based on the baseline feature extraction network. Then, for the purpose of accurate target localization and arbitrary-oriented ship detection, a rotation region proposal network and skew non-maximum suppression are employed. Finally, on account of the disadvantages that the employment of a rotation bounding box usually causes more false alarms, we implement inclined context feature pooling on rotation region proposals. A dataset including port images collected from Google Earth and a public ship dataset HRSC2016 are employed in our experiments to test the proposed method. Experimental results of model analysis validate the contribution of each module mentioned above, and contrast results show that our proposed pipeline is able to achieve state-of-the-art performance of arbitrary-oriented inshore ship detection.