Can video cameras replace visual estrus detection in dairy cows?

A 6-mo experiment was conducted in a dairy herd to evaluate a video system for estrus detection. From October 2007 to April 2008, 35 dairy cows of three breeds that ranged in age from 2 to 6 yr were included in the study. Four daylight cameras were set up in two free stalls with straw litter and connected to a computer equipped with specific software to detect movement. This system allowed the continuous observation of the cows as well as video storage. An observation method related to the functionality of the video management software ("Camera-Icons" method) was used to detect the standing mount position and was compared to direct visual observation (direct visual method). Both methods were based on the visualization of standing mount position. A group of profile photos consisting of the full face, left side, right side, and back of each cow was used to identify animals on the videos. Milk progesterone profiles allowed the determination of ovulatory periods (reference method), and a total of 84 ovulatory periods were used. Data obtained by direct visual estrus detection were used as a control. Excluding the first postpartum ovulatory periods, the "Camera-Icons" method allowed the detection of 80% of the ovulatory periods versus 68.6% with the direct visual method (control) (P = 0.07). Consequently, the "Camera-Icons" method gave at least similar results to the direct visual method. When combining the two methods, the detection rate was 88.6%, which was significantly higher than the detection rate allowed by the direct visual method (P < 0.0005). Eight to 32 min (mean 20 min) were used daily to analyze stored images. When compared with the 40 min (four periods of 10 min) dedicated to the direct visual method, we conclude that the video survey system not only saved time but also can replace direct visual estrus detection. (C) 2012 Elsevier Inc. All rights reserved.