Under the banner of internet of things (IOT),intelligent video survei llance is an emerging research direction in the field of computer vision.It focuses on how to describe,analyze and unders tand the contents of the surveillance video by using image and video processing technology and artificial intelligence techniqu es.Then it can control the video surveillance system according to the analysis results,making the video surveillance system h ave a higher level of intelligence.Moving object detection is an important part in intelligent video surveillance of IOT.B ut there are drawbacks on the illumination change,complex background and shadow in the traditional methods.B ased on the variety of shadow detection methods,we find that only using one feature can not obtain the result of det ection accurately.Then we present a new method of shadow detection for moving object based on machine learning,which co ntains color information,the invariance of optical and texture features.Through machine learning,it can adaptively do s hadow detection for moving object.Based on the comprehensive analysis of the detecting results of three kinds of informatio n,the shadow can be effectively determined. The method has combined the advantages of various methods,and it can get ideal e ffect and efficiency in the application of IOT.