Based on the physical requirements of the senior high school entrance examination in Shenzhen, Zhuhai, Zhongshan and other cities in Guangdong Province, and with reference to the National Student Physical Health Standard and the Compulsory Education Physical Education and Health Curriculum Standard (2023 Edition), a set of intelligent training and examination system composed of training and examination robots and server-side big data platform is designed. At the beginning of the assessment, the training robot verifies the identity of the participants through face recognition or swipe card. After successful authentication, the robot can use the camera to identify the bone of the examiner and classify the posture according to the selected sports category. Judge whether the appraiser's actions are up to standard, count the actions up to standard, prompt invalid non-standard actions, and correct the deviation in the form of dialogue through robots. After the assessment is completed, the collected data will be uploaded to the big data platform for users to query the results and analyze the training situation, and assist students to exercise more accurately. The key technology of this project is bone recognition, posture classification, standard action comparison, video remote inspection and so on. Firstly, the three-dimensional bone data of human motion is obtained from the depth camera, and the data is standardized. Secondly, the data is input into the action capsule model, and the output vector of each action category is compared with the standard technical action library by using the classifier, so as to determine the final action category. Finally, the user is corrected and counted. In the early preparation of this project, the prototype of the product has been built, and five platforms of big data on the server side have been successfully built, which have realized information management, attitude data detection, training and examination data analysis, physical fitness data statistics and network security guarantee respectively. In addition, the project adopts the separation of front and back ends, and has realized the perfect data interface design and business function design. The multi-terminal linkage mode is adopted to provide the school with a complete set of robot equipment for daily assessment and military training, and the school management is provided with general statistical data of all grades and classes in the background in a visual way. A big data management platform is provided for school teachers and coaches, which can count and update students' physical fitness in real time. It provides posture knowledge application software and applets for students and parents, at ordinary times, mobile phones can be used for home training anytime and anywhere. Young Xiong-a smart training and examination system based on human posture classification, is committed to promoting technological innovation, model innovation and industrial innovation in the smart training industry in the new era. The corresponding national "14th Five-Year" Sports Development Plan provides all primary and secondary school students with a full set of integrated "smart training", "smart assessment", "personalized training", "big data analysis" and other exercise programs.