Research On Human Body Keypoint Detection System Oriented To Shelf Product Concern

The flourishing development of AI technology has made some industries move to a new stage,especially in emerging industries such as autonomous driving,security monitoring,face recognition,and smart retail,which have achieved advanced development and applications.In the smart retail industry,the analysis of customers' stopping points and other information can enable businesses to obtain more comprehensive user portraits,so as to adjust sales strategies,improve customer satisfaction and maximize profits.Therefore,in the process of customer behavior analysis,fast and accurate pedestrian detection and extraction of key points of the human body have become necessary links.This paper mainly studies the human body keypoint detection system in the environment of shelf goods,so that the system can detect human body key points quickly and accurately.The main work of this article is as follows:1.Research on data enhancement methods based on depth images.Based on the imaging principle of depth images,a data enhancement method based on depth images is proposed,which can expand the depth image data during model training and improve the generalization ability of the training model results.2.Research on Optimization Method of Network Structure Model.Study several existing deep learning-based network structure model optimization methods and combine several methods to propose a lightweight and efficient feature extraction network architecture.This model architecture reduces the amount of model parameters and calculation At the same time,the accuracy of the model is improved.3.Research on human bone keypoint detection algorithm.Aiming at the real-time and accuracy requirements of human key point detection in the business-oriented supermarket scene,a human keypoint detection algorithm based on depth images is proposed,and has good generalization ability.4.Experimental study of human body keypoint detection system for shelf product environment.In order to verify the performance of the detection system,an accuracy experiment was performed on the detection algorithm under the commercial supermarket shelf scene dataset.A real-time comparison experiment with mainstream human keypoint detection algorithms was performed.The generalization ability of the detection algorithm is verified in different supermarket shelves scenarios.