An Agile and Efficient Neural Network Based on Knowledge Distillation for Scene Text Detection

Table 6

Comparison results of the parameters size. The smallest size is highlighted in italic.

Method	Backbone	Size
PSENet [16]	ResNet50	27.3 M
TextSnake [15]	VGG16	18.2 M
CRAFT [28]	VGG16	19.8 M
DBNet [1]	ResNet18	13.2 M
The proposed (Pr0.5)	ResNet18	6.8 M