Towards precision bovine insemination using artificial intelligence
In the cattle breeding sector, detecting the heat period in a cow is a prerequisite for successful artificial insemination and conception. Until now, the detection was done by visual observations of behavioral signs by a human operator who could be assisted by automated analysis systems. An original study combining artificial intelligence and innovative insemination equipment has shown that it is possible to detect heat quickly and more efficiently by endoscopic analysis of the cow’s genital tract.
Bovine artificial insemination is a biotechnology that fully contributes to the development of sustainable agriculture thanks to the advantages it offers in terms of health safety, genetic gain and economic costs. The main condition for a successful insemination of a cow (conception) is the detection of her heat period.
Typically, this period is identified by the farmer through visual observation of a set of behavioral signs exhibited by the cow. However, using this method of detection in large herds is a tedious process. In addition, cow signs may be affected by various health and environmental factors, which reduces the effectiveness of detection.
To address these limitations, a novel endoscopic heat detection system was developed using an innovative insemination technology called Eye Breed and an artificial intelligence model designed and optimized for smartphone deployment.
As illustrated in Figure 1, the Eye Breed device, equipped with an on-board camera and connected to a smartphone, is inserted into the cow’s genital tract by a human operator who supervises a simulated insemination operation in real time. The recorded video is then automatically analyzed by the artificial intelligence model embedded in the smartphone to determine the heat status of the cow in less than 20 seconds. The artificial intelligence model developed employs a deep learning technique based on Convolutional Neural Networks (CNN), whose structure is inspired by the receptive field structures found in the human primary visual cortex.
The developed system has shown a high performance on the 32 cows tested so far with a good detection rate of 87.5%. In the medium term, the system will be deployed and tested on several farms in the northern region of France. Also, new avenues of research will be studied in connection with the development of intelligent systems to help diagnose cow pathologies by using the Eye Breed device.
This collaborative work between several research teams from IEMN (BioMEMS and COMNUM groups), JUNIA, Gènes Diffusion, ELEXINN and LIMMS, is part of a FEDER research project of the Hauts-de-France region. The approach developed and the results obtained have been published in the reference .
https://doi.org/10.1007/s10489-021-02910-5 He, R., Benhabiles, H., Windal, F., et al. A CNN-based methodology for cow heat analysis from endoscopic images. Applied Intelligence – The International Journal of Research on Intelligent Systems for Real Life Complex Problems – SPRINGER – published online 27 October 2021.