<p>Ancient murals are a precious human cultural heritage. They are suffering from various diseases due to long-term environmental exposure and human intervention. Accurate detection of the diseased areas in ancient murals is of great research value. This paper proposes a dual-encoder hierarchical feature fusion (DEHFF) network based on convolutional neural networks (CNNs) and a Transformer. In this network, a dual encoder (DE) is proposed to extract local features and global context information simultaneously. It can enhance the feature representation capability of the network. Moreover, a hierarchical feature fusion (HFF) module is introduced to facilitate multi-level feature fusion. A contextual channel attention (CCA) module is designed to capture the dependencies across feature channels for the bottleneck layer. We conduct experiments on the Mogao Grottoes murals of Dunhuang and the Luose Temple murals in Yunnan. Experimental results show the proposed network outperforms other state-of-the-art approaches when evaluated by visual comparison and objective metrics.</p>