Real-time precise mood assessment of Household and Farm Animals (with or without CBD)

Forkosh Oren, HUJI, Faculty of Agricultural, Food and Environmental Quality Sciences

While there is an upward trend for animal welfare, issues are only addressed when animals are sick, injured, very lean, dirty, aggressive, or show avoidance behavior. For prey animals like cows and cats, these behaviors are only exhibited after a prolonged period, making recovery nearly impossible. Tracking positive welfare can predetermine problems before they surface, prevent escalation, and provide a better quality of life for the animals in general. Yet, the challenge is that positive welfare is often hard to measure.

Our Innovation
We are developing the methodology to systematize an animal’s positive welfare based on natural behaviors utilizing machine learning. This continuous automatic and objective platform will allow owners to prophylactically intervene. The animal’s mood is monitored using a simple electronic tag taking various measurements such as movement, position, respiratory rate, and body temperature. Any change in the normal behavioral or physiological patterns associated with anxiety (such as an elevated respiratory rate without increased activity or decreased appetite) can signal the need for intervention. The Invention is a system that combines novel yet off-the-shelf tracking tags with our Artificial Intelligence-based algorithms and software.

  • Significant Advantages of Our System:

    Individualized approach: We track each animal's behavioral fingerprint and personality to detect abnormalities (Forkosh et al., Nature Neuroscience et al., 2019[1])
  • Complex behaviors: Our technique can identify complex behaviors involving social interactions, hierarchy, and spatial preferences. Some of these behaviors are predefined (such as agonistic interactions), and some are learned by the algorithm (Forkosh et al., Patterns Cell-Press 2021[2])
  • Based on off-the-shelf hardware: Simplifying the development process and focusing on analysis.
  • Cloud-based solution: The analysis can either be done onsite or on the cloud.

We propose a mathematical approach that directly measures the animal’s behavior for positive welfare. Our algorithm will use a high-dimensional complex behavioral space obtained from movement trajectories, posture, and physiology (similar to Shemesh & Forkosh et al., Nature Neuroscience, 2016[3]). Currently, we have a working pilot setup at the Agricultural Research Organization (ARO; Volcani Center). We are constantly tracking the location, posture, and social behavior of a small group of six cows using a combination of cameras, depth cameras, and position tags. We use this data to evaluate the unique behavioral fingerprint of each cow and compare it, among others, to its welfare, productivity, and various endocrinological readings. An additional system capable of simultaneously tracking over 80 animals for prolonged periods is under construction at the Eshel HaNasi village.

The proposed approach provides the most advanced system to track dairy cows' health, welfare, and reproductive cycle. The system will deliver early alerts to offset health and welfare issues as well as offer solutions, while preventing further escalation. This system will benefit the animals welfare and health as well as productivity and cost for the owners.

[1] https://www.nature.com/articles/s41593-019-0516-y

[2] https://www.sciencedirect.com/science/article/pii/S2666389920302646

[3] https://www.nature.com/articles/nn.4346

Contact for more information:

Ilya Pittel
Contact ME:
All projects by:Forkosh Oren (1)