Sports Video Classification and Labeling Algorithm Based on Data Mining Technology

With the explosive development of video media, sports workers have accumulated rich experience in sports competitions, training and technical analysis videos. These data contain rich sports information and personal characteristics of athletes. In these videos, how can sports workers quickly find the video materials they need? The traditional text recognition method cannot meet the demand because of its slow speed. If the conventional text recognition method is used, it is equivalent to looking for a needle in a haystack. Due to the large amount of data in sports video, the traditional text recognition method is difficult to meet people’s high demand for real-time and accuracy. Today, with the vigorous development of data mining (DM for short here) technology, using DM to classify and label a large number of sports videos can well solve the problem of low efficiency of traditional text recognition video recognition. At the same time, the trained classifier can help users find the information and action patterns of interest. Therefore, based on the support vector machine (SVM) algorithm, this paper studies the classification of moving video. Finally, take the video of ice and snow sports events in the past three years as the data sample, the experimental platform uses Windows 7 system PC, the simulation environment is Matlab 2012, and then use OpenCV (Open-Source Computer Vision Library) to process the video. At the same time, by comparing the advantages and disadvantages of different classifiers, the most suitable classifier for this study is selected. Finally, this paper applies the trained model to the analysis of the actual collected ice and snow game video. The experimental results show that the average recall and precision of the SVM algorithm are 0.9527425 and 0.945192, respectively. It can be seen that SVM algorithm can achieve the purpose of processing complexity, and a large number of sports videos have been effectively managed and classified rapidly, with high classification accuracy.