Research on Pedestrian Detection Algorithm Based on MobileNet-YoLo

Table 1

Symbol table.

A	Input image size
B1 and B2	Prediction box size
n	Number of bounding boxes per grid prediction
N	Number of detected categories
	Total loss
	Overlap area
	Distance
	Aspect ratio
and	Prediction box
and	The center point of the box and box
	The diagonal length of the frame
	Euclidean distance
and	The width of box and box
and	The height of box and box
	Weighing parameters
Center	Center of all clusters
Box	Sample clustering results
IoU	The intersection ratio of all centers to all boxes
mAP	Average precision means
Speed	Transfer frames per second
Params	A total number of participants
AP	Average accuracy
i	A category
P	Accuracy
R	Recall rate
	Mapping between precision and recall
TP	Several samples for which both the detection category and the true label are i
FP	Several samples with detection category i and true label not i
FN	Detect the number of samples with category not i but with true label i
, , and	Loss function