VSCnet: A View Sematic Contexutal Based Network for Multi-label Classification of Cell

Medical image analysis is one of the most important applications of computer vision. Because of the complexity of medical images, classifiers need not only efficient feature extractors, but also information of different of vision semantic context to make prediction. As a result, many common deep learning models do not perform well in medical images. In this paper, we propose a novel architecture based on multi-view learning, which generate the view of label semantics context and view of global spatial context to make supplement each other’s blind spots. This method archive state of art performance in our dataset.