Puzzle: taking livestock tracking to the next level.

Animal behavior is a critical aspect for a better understanding and management of animal health and welfare. The combination of cameras with artificial intelligence holds significant potential, particularly as it eliminates the need to handle animals and allows for the simultaneous measurement of various traits, including activity, space utilization, and inter-individual distance. The primary challenge in using these techniques is dealing with the individualization of data, known as the multiple object tracking problem in computer science. In this article, we propose an original solution called "Puzzle." Similar to solving a puzzle, where you start with the border pieces that are easy to position, our approach involves commencing with video sequences where tracking is straightforward. This initial phase aims to train a Convolutional Neural Network (CNN) capable of deriving the appearance clues of each animal. The CNN is then used on the entire video, together with distance-based metrics, in order to associate detections and animal id. We illustrated our method in the context of outdoor goat tracking, achieving a high percentage of good tracking, exceeding 90%. We discussed the impact of different criteria used for animal ID association, considering whether they are based solely on location, appearance, or a combination of both. Our findings indicate that, by adopting the puzzle paradigm and tailoring the appearance CNN to the specific video, relying solely on appearance can yield satisfactory results. Finally, we explored the influence of tracking efficacy on two behavioral studies, estimating space utilization and activity. The results demonstrated that the estimation error remained below 10%. The code is entirely open-source and extensively documented. Additionally, it is linked to a data-paper to facilitate the training of any automatic detection algorithm for goats, with the goal of fostering open access within the deep-learning livestock community.

Primiparous sow behaviour on the day of farrowing as one of the primary contributors to the growth of piglets in early lactation.

Large White and Meishan sows differ in maternal ability and early piglet growth. We investigated the relationships between 100 maternal traits, grouped into 11 blocks according to the biological function they describe and litter growth over three successive periods after birth (D0-D1, D1-D3 and D3-D7; D0 starting at the onset of farrowing), as a measure of sow investment in early piglet production. Within- and between-breed variation was exploited to cover a maximum of the variability existing in pig maternal populations. The objective was to quantify the contribution of maternal traits, including functional traits and behavioural traits, to early litter growth. Multivariate analyses were used to depict correlations among traits. A partial least square multiblock analysis allowed quantifying the effect of maternal traits on early growth traits. Partial triadic analyses highlighted how sow behaviour changed with days, and whether it resulted in changes in litter growth. Several behavioural traits (standing activity, reactivity to different stimuli, postural activity) and functional traits (body reserves, udder quality) at farrowing contributed substantially to litter growth from D0 to D7. Sow aggression towards piglets and time spent standing at D0 were unfavourably correlated to D1-D3 litter growth. Time spent lying with udder exposed at D0 was favourably correlated to D1-D3 litter growth. The farrowing duration was negatively correlated to D0-D1 and D1-D3 litter growth. Furthermore, D3-D7 litter growth was positively correlated to feed intake in the same period. Several behavioural traits and some functional traits influence early litter growth. The contribution of sow behaviour was greater in the critical period around farrowing than in later days.

CherryChèvre: A fine-grained dataset for goat detection in natural environments.

We introduce a new dataset for goat detection that contains 6160 annotated images captured under varying environmental conditions. The dataset is intended for developing machine learning algorithms for goat detection, with applications in precision agriculture, animal welfare, behaviour analysis, and animal husbandry. The annotations were performed by expert in computer vision, ensuring high accuracy and consistency. The dataset is publicly available and can be used as a benchmark for evaluating existing algorithms. This dataset advances research in computer vision for agriculture.

Analysis of image-based sow activity patterns reveals several associations with piglet survival and early growth.

An activity pattern describes variations in activities over time. The objectives of this study are to automatically predict sow activity from computer vision over 11 days peripartum and estimate how sow behavior influences piglet's performance during early lactation. The analysis of video images used the convolutional neural network (CNN) YOLO for sow detection and posture classification of 21 Large White and 22 Meishan primiparous sows housed in individual farrowing pens. A longitudinal analysis and a clustering method were combined to identify groups of sows with a similar activity pattern. Traits under study are as follows: (i) the distribution of time spent daily in different postures and (ii) different activities while standing. Six postures were included along with three classes of standing activities, i.e., eating, drinking, and other, which can be in motion or not and root-pawing or not. They correspond to a postural budget and a standing-activity budget. Groups of sows with similar changes in their budget over the period (D-3 to D-1; D0 and D1-D7) were identified with the k-means clustering method. Next, behavioral traits (time spent daily in each posture, frequency of postural changes) were used as explanatory variables in the Cox proportional hazards model for survival and in the linear model for growth. Piglet survival was influenced by sow behavior on D-1 and during the period D1-D7. Piglets born from sows that were standing and doing an activity other than drinking and eating on D-1 had a 26% lower risk of dying than other piglets. Those born from sows that changed posture more frequently on D1-D7 had a 44% lower risk of dying. The number of postural changes, which illustrate sow restlessness, influenced piglet growth in the three periods. The average daily gain of piglets born from sows that were more restless on D1-D7 and that changed posture more frequently to hide their udder on D0 decreased by 22 and 45 g/d, respectively. Conversely, those born from sows that changed posture more frequently to hide their udder during the period of D1-D7 grew faster (+71 g/d) than the other piglets. Sow restlessness at different time periods influenced piglet performance.

A Non-Invasive Millimetre-Wave Radar Sensor for Automated Behavioural Tracking in Precision Farming-Application to Sheep Husbandry.

The automated quantification of the behaviour of freely moving animals is increasingly needed in applied ethology. State-of-the-art approaches often require tags to identify animals, high computational power for data collection and processing, and are sensitive to environmental conditions, which limits their large-scale utilization, for instance in genetic selection programs of animal breeding. Here we introduce a new automated tracking system based on millimetre-wave radars for real time robust and high precision monitoring of untagged animals. In contrast to conventional video tracking systems, radar tracking requires low processing power, is independent on light variations and has more accurate estimations of animal positions due to a lower misdetection rate. To validate our approach, we monitored the movements of 58 sheep in a standard indoor behavioural test used for assessing social motivation. We derived new estimators from the radar data that can be used to improve the behavioural phenotyping of the sheep. We then showed how radars can be used for movement tracking at larger spatial scales, in the field, by adjusting operating frequency and radiated electromagnetic power. Millimetre-wave radars thus hold considerable promises precision farming through high-throughput recording of the behaviour of untagged animals in different types of environments.

Mixed grazing of adult goats and cattle: Lessons from long-term monitoring.

Mixed grazing of breeding goats and cattle (goats to cattle ratio: about 50 %, based on metabolic weight) was monitored for 2 years on a rotational pasture with the two species grazing together, then for 5 years with cattle grazing immediately after goats. For both modalities, the level of goat parasite infection was not significantly different from that of the control groups. Nevertheless, the association allowed a slight improvement in kid growth and goat productivity, probably in relation to a better food quality. The response of adult goats to mixed grazing is therefore very different from that previously obtained with kids post-weaning. The question of the relationship between heterogeneity of pastures, knowledge of their environment, grazing behaviour of adult goats and risk of infection with gastrointestinal nematodes requires further investigation.

Goats worm burden variability also results from non-homogeneous larval intake.

For small ruminants, Gastrointestinal Nematodes (GINs) are responsible for severe economic losses and they are also an animal welfare problem. GIN use their host to reproduce and disperse eggs on the pasture, from where they can re-infect another animal. The high density of hosts on the pasture and the extreme tolerance of GIN to environmental constraints make GIN eradication almost impossible. In addition, significant resistance to anthelmintic treatment requires sustainable and integrated management to maintain the health and financial well-being of livestock farming. In this context, models of the complex interactions between host, GIN and environment can help us to design long term optimal management strategies. To build such models, quantitative information is needed but are generally very challenging to collect. In this article, we focus on the number of ingested larvae per animal, which we propose to characterise by using a simulation framework based on the estimation of the spatial distribution of the host over time. Our framework allows us to show that worm burden individual variation is not only explained by the host's genetics, as is often the case, but is also a result of the grazing spatial process.

Dynamic minimum set problem for reserve design: Heuristic solutions for large problems.

Conversion of wild habitats to human dominated landscape is a major cause of biodiversity loss. An approach to mitigate the impact of habitat loss consists of designating reserves where habitat is preserved and managed. Determining the most valuable areas to preserve in a landscape is called the reserve design problem. There exists several possible formulations of the reserve design problem, depending on the objectives and the constraints. In this article, we considered the dynamic problem of designing a reserve that contains a desired area of several key habitats. The dynamic case implies that the reserve cannot be designed in one time step, due to budget constraints, and that habitats can be lost before they are reserved, due for example to climate change or human development. We proposed two heuristics strategies that can be used to select sites to reserve each year for large reserve design problem. The first heuristic is a combination of the Marxan and site-ordering algorithms and the second heuristic is an augmented version of the common naive myopic heuristic. We evaluated the strategies on several simulated examples and showed that the augmented greedy heuristic is particularly interesting when some of the habitats to protect are particularly threatened and/or the compactness of the network is accounted for.