Mixed context networks for semantic segmentation

摘要: Semantic segmentation is challenging as it requires both object-level information and pixel-level accuracy. Recently, FCN-based systems gained great improvement in this area. Unlike classification networks, combining features of different layers plays an important role these dense prediction models, contains levels. A number models have been proposed to show how use features. However, what the best architecture make still a question. In paper, we propose module, called mixed context network, that our presented system outperforms most existing semantic by making module.