When raising livestock, feeding and associated behaviors need to be quantified accurately in order to detect any health and welfare problems at an early stage, according to information posted on the European Commission's CORDIS website for European Union research results.
Changes in feeding behaviors are a sign of such problems, and even subtle differences in the way an animal consumes its food could help in spotting health and welfare issues in livestock. Researchers supported by the EU-funded HealthyLivestock and Feed-a-Gene projects have developed a promising new method for monitoring pig feeding and foraging that could help with the early detection of such problems.
Described in a paper published in Biosystems Engineering, the automated detection method can be used in a variety of husbandry and management situations, the researchers said. Based on convolutional neural networks, the two-dimensional camera-based deep learning method automatically detects pig feeding behavior without the use of additional sensors or individual marking.
According to the study, “the system operates on grayscale video images and was trained to handle the constantly changing farm conditions, e.g., lighting conditions, problems of occlusion caused by other pigs and insects occluding the image from the camera.”
Feeding behaviors are not estimated using traditional pig tracking methods; instead, the researchers used “GoogLeNet-like architectures ... to monitor a smaller, predefined pen area covering two food troughs and a simple, clearly defined area in front of those troughs. In this way, the proposed system avoids short [identification] track-related issues, which can continuously distort the accumulative feeding behavior recognition process.”
Detecting subtle behavioral changes
Detection of feeding behavior is fast (0.02 seconds per image) and accurate (99.4%), the researchers said. Unlike with traditional pig tracking, the system doesn’t overestimate the actual time spent feeding. This is because it can distinguish between non-nutritive visits (NNVs) to the feeding area (where the feet but not the head are inside the feeding trough) and actual feeding (with the head also inside the trough).
“As our system focuses only on a subset of available feeding troughs within a commercial context, we demonstrate that sufficient data can be collected from this subset to identify changes associated in feeding behaviors at group level,” the study found.
The method was first validated using video footage from a commercial pig farm in different settings. Next, during a planned period of food restriction in which the pigs received 80% of their daily food for four consecutive days, the team tested the method’s ability to detect changes in feeding and NNV behaviors.
“We found that the method was able to automatically quantify the expected changes in both feeding and NNV behaviors,” the researchers said.
Furthering the aims of HealthyLivestock (Tackling Antimicrobial Resistance through Improved Livestock Health & Welfare) and Feed-a-Gene (adapting the feed, the animal and the feeding techniques to improve the efficiency and sustainability of monogastric livestock production systems), the method could help in the early detection of health and welfare challenges of commercial pigs, the researchers said. Feed-a-Gene ended in early 2020, while the four-year HealthyLivestock project concludes in 2022.