Brand LOGO Image Recognition and Inquiry Creative Teaching Design Based on Deep Learning

Table 2

Results of different algorithms.

Algorithm (model)	mAP (%)	Total train time (min)	Test time (ms)/image	Hyperparameters
YOLO algorithm	32.7	706	22	mp, , 0.2
R-CNN algorithm	50.9	1285	25685	Pixels of the image: , , and; Convolution kernel: ,; Step size: 1; The number of convolution kernels: 64
Faster R-CNN algorithm	56.9	822	167	The number of convolution kernels: 128; : 1 ( controls the balance between the two task losses) RoI (region proposal): 64
ContextNet algorithm	62.5	1058	198	Stride: 16; Backbone: Xception65