Due to the application of internet of things (IoT) to largescale cow breeding, mass of multiscale data and multidivisional sensor data and video monitoring data of cow individuals were collected. Therefore, it is significant to dig out useful information about features of healthy reproduction behavior for development of scientific largescale breeding measures and improve economic benefits from cow breeding. For the rapid and accurate identification of cow reproduction and healthy behavior from mass surveillance video, totally 400 head of young cows and lactating cows were taken as the research object and cow behavior from the dairy activity area and milk hall ramp was analyzed. The method of object recognition based on image entropy was proposed, aiming at the identification of motional cow object behavior against a complex background. Calculation of a minimum bounding box and contour mapping was used for the realtime capture of rutting span behavior and hoof or back characteristics. Then, by combining the continuous image characteristics with movement of cows for 7d, abnormal behavior of dairy cows from healthy reproduction can be quickly distinguished by the method, which improved the accuracy of the identification of dairy cows characteristics. Cow behavior recognition based on image analysis and activities was proposed to capture abnormal behavior that had harmful effects on healthy reproduction and improve the accuracy of cow behavior identification. The experimental results showed that through target detection, classification and recognition, the recognition rates of hoof disease and heat in the reproduction and health of dairy cows were greater than 80%, and the false negative rates of oestrus and hoof disease reached 3.28% and 5.32%, respectively. This method can enhance the real -time monitoring of cows, save time and improve the management efficiency of large scale farming.