Pig-rearing can rely on artificial intelligence to tackle its challenges

Production sites tend to expand

Most of the breeders who stop their activity sell their holdings to increase neighbouring production sites. Holdings are becoming larger and larger and multi-site. The geographical extensions bring about side-effects (management, distance, follow-up) and at the same time, breeders try to cut down labour costs. In that context, remote follow-up solutions are becoming necessary to better support breeders and their employees in monitoring their holdings.

Expectations around breeding conditions

In pig holdings, ambient atmosphere quality is a vital parameter which affects breeding performances. But atmosphere parameters (gas levels (NH3, Co2), humidity, air flow rate, luminosity) are not always known. In an approach to use less or no medication, a good knowledge of ambient atmosphere enables the breeder to follow up all the indicators so as to avoid health problems while improving performances (growth, decrease in mortality rate, nutrition, etc…)

Artificial intelligence to meet tomorrow’s issues

Strong evolutions are expected in pig holdings especially concerning animals, behaviour and activity. Moreover, breeders have to know whether the implementation of those new technologies is relevant. In order to implement and assess those evolutions gathering objective data is necessary. Those data will have to be quickly analyzed so that they can be adjusted and deployed in all kinds of holdings (pigs, broilers, laying hens,…).

Pictures, videos, data coming from sensors, declarative data have to be consolidated, sorted out and aggregated to make it possible to deduce follow-up indicators for tomorrow’s holdings. Image and data analysis tools which use a deep-learning approach based on artificial intelligence algorithms will be used to set-up indicators and then build-up predictive models (behaviour, health, welfare, nutrition).

“Peek and Sense” solution

Peek and Sense solution is based on a connected box called “Peek” : it collects pictures and videos of animals by using up to three cameras. These images and videos are coupled with atmosphere data that have been collected closest to animals (temperature, humidity, Co2, air flow rate, etc…). A daily precise follow up can then be implemented in the production site and as a consequence, movements in and out of the building (a key issue in health safety) can then be drastically limited, while optimizing trips and interventions. The measured values are used as indicators which generate notifications and alerts. The volume of collected data makes it possible to characterize breeding conditions in order to improve them and take part in food traceability through injecting them into blockchains.

Julie Champion,

Business developer at COPEEKS