Currently mainly static, “resource based” indicators dominate certificates and labels of animal health and welfare, e.g. how many kg live weight /m² or how much daily weight gain is allowed. There is a common understanding those KPIs are only second best choice, because they do not actually look at specific animals or flocks. Yes, they are pragmatically easy to audit by independents but an “output based” approach is needed. This means, that detecting how actually a flock and birds are behaving is more important than general limits and figures. Cameras, real-time image recognition and data analytics have huge potential to monitor and improve specifically welfare. This is reflected by recently
awarded research grants by the Foundation for Food and Agriculture Research (FFAR) and McDonald’s (one of the largest poultry meat buyer). So, the race is open to contribute to more surveying broilers and contribute to more animal welfare specifically. To our opinion cameras and images are one important new tool, but for sure results need to be smartly combined with other factors like feed composition, farm management and health issues to build up real insight - just like an experienced and good farmer does today.
To bring together the expertise of the best for the sake of improving the industry was the one reason why Evonik decided to build up its Porphyrio Precision Farming Platform for poultry. We think, that although the general understanding about the requirements of chicken has increased over the last decades and farming practices generally have been professionalized, there is certainly a lot of potential to improve. Better surveying the flocks in order to detect any anomaly as early as possible would allow for earlier interventions and ideally avoiding medical treatments. Furthermore: while some experienced farmers might spot any change already today by just watching, listening and looking at birds and data, the less skilled farmers might not.
Some early indicators for health issues of chicken are movement patterns, low populated spots in the barn, too fast or too slow growth. Technology now offers tools for improvement, supporting the farmer to look after his flock when he or she is not in the house. Automated analysis of top-down pictures from the barn should allow us to detect welfare and health issues early and notify farmers to take action in case of an anomaly. One parameter is the amount of healthy movement inside the chicken flock. Exceptionally low or high movement can be an early indicator for uncomfortable temperatures, cold spots due to wrong ventilation, a lack of feed or water in certain areas of the house or even infections that might need medical treatment. Currently, daily growth is mainly planned and predicted based on a farmer’s experience, maybe by comparing with “standard growth curves” provided for the particular breed. Manual weighing of a rather small sample of chickens takes place regularly but induces stress to the animals and is labor intensive. Automatic scales exist but currently are limited in accuracy and only work in certain time spans within the growth cycle reliably. Additionally, there is no control of the distribution of chickens actually being measured by automated scales and no control about the representativeness of the chicken passing the scale for the whole flock.
Ideally, a visual measurement would take place using cameras, measuring a large sample of birds and providing a weight distribution estimation. Advantages of an optical measurement would be: No stress for the individual animal and it could be applied to hundreds or even thousands of birds at high frequency. Automated detection of offsets to the „standard growth curve“ would allow farmers to take action early. One of the practical challenges is that the birds flare up their feathers more or less in certain conditions and change their size significantly.
You will receive a large dataset of top-down pictures from barns for training purposes, expected daily growth curves and final chicken weights and distributions. The image-portfolio covers 24/7 observation in multiple runs which cover all flock-ages. Images are tagged with date and time. Actual average weights are given per date and respectively flock-age.
If the startup provides a successful implementation of the challenge, Evonik can provide future market access via its Porphyrio platform. This platform designed to offer digital services and data exchange within the poultry industry might be an ideal sales channel to interested farmers. Furthermore, Evonik can enable access to versatile datasets of different poultry houses which could be used to further evolve this use case.
A project duration of approximately 3 months is expected.