Classification of Tennis Video Types Based on Machine Learning Technology

With the rapid development of online video data, how to find required information has become an urgent problem be solved. This article focuses on sports videos and studies classification content-based retrieval techniques. Its purpose is establish a mark index content promote user acquisition through computer processing, analysis, understanding content. Video tennis high research application value. based selection basic frame each shot proposes algorithm for shots average grouping. Based this, we use color-coded spatial detection method detect type match. Then, it integrates results audiovisual analysis identify classify exciting events in matches. According statistics, although number people participating cannot enter top ten, spectators ranks fourth. Four tournaments, masters, crown tournaments are held every year around world. Watching large-scale international matches pillar leisure vacation many people. Tennis last from two hours four or more, there countless large small world year, so records created staggering. And artificial intelligence technology rarely used (5%), but football reached 50%. Therefore, when dealing with such amount urgently need fast effective information. The experiment machine learning proves that accuracy reaches 98%, this system feasibility.