Fig. 1From: Mural classification model based on high- and low-level vision fusionPerformance comparison for two types of convolution kernels: a 3 × 3 filter; b 5 × 5 filterBack to article page