To solve the problem of low detection and recognition rate of small-sized pedestrian objects in images, a pedestrian detection algorithm based on the improved multi-scale feature fusion is proposed.Firstly, the BN layer is fused into the convolution layer based on the original YOLOv3 model.Secondly, a detection layer is added, and the features of the high and low layers are combined and predicted by referring to the idea of feature pyramid.Finally, K-means clustering algorithm based on linear scaling is used to optimize the anchor box, so as to improve the detection effect of small-sized pedestrian.The experimental results on INRIA pedestrian dataset show that the accuracy of the improved algorithm reaches 91.4%, which is 3.4% higher than that of the YOLOv3 algorithm.The effectiveness of the proposed algorithm is also proved in complex monitoring environment.