A system and a method for the determination of animal behaviour during a stunning process

Abstract

The present invention is directed to a system and a method for measuring animal behavior during a stunning process in a testing or research situation, at a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method, which system and method is based on the use of the animals' activity, as observed using vision means, and as determined by a processing means configured for running an artificial intelligence algorithm, and which behavioral activity is used as an expression of the animals reaction to the given conditions.

Description

[0001] A SYSTEM AND A METHOD FOR THE DETERMINATION OF ANIMAL BEHAVIOUR DURING A STUNNING PROCESS

[0002] TECHNICAL FIELD

[0003] The present invention is directed to a system and a method for measuring animal behavior during a stunning process in a testing or research situation, at a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method, which system and method is based on the use of the animals' activity, as observed using vision means, and as determined by a processing means configured for running an artificial intelligence algorithm, and which behavioral activity is used as an expression of the animals reaction to the given conditions.

[0004] BACKGROUND ART

[0005] Gas stunning animals in a group is used in different situations. It is, e.g., used in slaughterhouses to stun pigs and poultry prior to killing. It can also, under certain conditions, be used to kill animals by prolonged exposure, which do not necessarily take place at a slaughterhouse (Council Regulation (EC) No 1099 / 2009, 2009 on the protection of animals at the time of killing). Furthermore, it can also be used to stun and kill rodents that are bred to, and used as laboratory animals, e.g., in the pharmaceutical industry (Directive 2010 / 63 / EU, 2010 on the protection of animals used for Scientific Purposes).

[0006] The method described herein is tested on pigs stunned at a commercial slaughterhouse with CO2. However, the method will be applicable in general to assess animals that are stunned by a gas method or killed by prolonged exposure to a stunning gas. This with the aim of monitoring and documenting how the animals react to the given conditions during the stunning process. Thus, it can be used as a tool to make continuous monitoring of the animals to detect deviations and to become aware that necessary measures must be taken. It can also be used in test-situations if you want to compare, for example, how animals react to different types of gas stunning or whether a given management measure has an impact on the stunning process. For example, does chancing the temperature result in less response to the stunning.

[0007] The practice of using controlled atmosphere stunning with carbon dioxide (CO2) to stun pigs at slaughter in abattoirs is the most applied method in commercial settings in Europe. The use of CO2 offers several advantages, such as management of pigs in groups with minimal human intervention, thereby reducing stress. Furthermore, it poses a low risk of errors by operators and ensures a reliable level of stunning effectiveness, and there is minimal risk of animals regaining consciousness. The method, however, causes concerns in relation to animal welfare due to some pigs showing aversive behavior on exposure to the gas. At the European level, a large effort is put on improving stunning methods, and on finding alternative and more animal welfare friendly methods for stunning pigs. In this respect, the effect of different improvements and different gas alternatives have been and are being tested currently.

[0008] When pigs are stunned with gas in commercial settings (currently only CO2), the pigs are typically placed in a cage, in jargon called a gondola, descending into an atmosphere of a high concentration of CO2, i.e., a concentration of about 80% or more. Gondolas exist in different sizes within different systems. But a common feature, when pigs are stunned under commercial settings, is that the pigs stand relatively close together in the gondola, thus representing a high stocking density.

[0009] Measuring animal welfare during stunning with CO2 (or another gas) under commercial conditions in the stunner is consequently challenging. The stocking density makes counting, recording and interpretation of animal behavior difficult. Nonetheless it is crucial, when trying to find alternative gas-stunning methods, or when trying to improve existing methods, that the comparison is done in an objective and repeatable manner that is maintainable under commercial conditions.

[0010] An unmet need exists for the provision of methods to examine how pigs respond to different stunning methods (improved / altered), and methods that in general can ascertain and document the behavior and / or welfare of the pigs, and the stunning quality, in a commercially used stunning system during the stunning procedure.

[0011] SUMMARY OF THE INVENTION

[0012] In known studies of stunning procedures, the animals' response is typically calculated by counting the occurrence and degree of specific behavioral patterns, e.g., escape behavior, jumping, clamming, or muscle cramps. The animals' individual reactions are counted and scaled, and an interpretation is made of the significance for, and influence on, the behavior and / or welfare of the animals is interpreted.

[0013] The system and method described herein assume that the behavior pattern of the animals' correlates to the increased movement typically recorded in this setting, which can be analyzed and determined by use of artificial intelligence to sum up the animal's activity.

[0014] The findings obtained by the system and the method of the invention have been manually verified by comparing recordings with deferent levels of activity (as determined by the described method) with what is observed on recordings. This verification confirms that the method is indeed useful to determine the effect on the pigs’ activity responses to different conditions during stunning. The system and method described herein relate to measuring the activity of the pigs during stunning for the purpose of determining stunning quality or welfare characteristics, based on automated measurements of animal activity or movements of the animals during the stunning process.

[0015] The system and method described herein can also be used to do general welfare monitoring of the animals in the stunner. For example, continuous surveillance to determine whether the animal's responses exceed a certain level, and / or whether there is a need for action.

[0016] Therefore, in its first aspect, the present invention provides a system for measuring the behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning method in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method.

[0017] The system of the invention may be characterized by comprising the following essential elements: a vision means (2), in communication with a processing means (3), which vision means (2) is configured to monitor, and to obtain video data, on the entire group of animals (1) subjected to the stunning process; and the processing means (3), in communication with the vision means (2), which processing means (3) is configured for collecting and computing data obtained from the visions means (2); in which system computing by the processing means (3) is accomplished using an artificial intelligence (Al) algorithm, to identify and sum up the activity of the animals, as observed by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals is compromised.

[0018] In another aspect, the invention provides a method for measuring behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning method in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method.

[0019] The method of the invention may be characterized by comprising the following subsequent steps:

[0020] (i) subjecting the group of animals (1) to be subjected to the stunning process, to monitoring by a vision means (2), configured to monitor all animals during the stunning process; and

[0021] (ii) processing the data, obtained according to step ii, in a processing means (3); in which method the processing means (3) is configured for running an artificial intelligence (Al) algorithm, and which processing means (3) is configured to identify and sum up the activity of the animals (1), as monitored by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals are compromised.

[0022] Other objects of the invention will be apparent to the person skilled in the art from reading the following detailed description and accompanying drawings.

[0023] Any combination of two or more of the embodiments described herein is considered within the scope of the present invention.

[0024] BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The present invention is further illustrated by reference to the accompanying drawing, in which:

[0026] Figs. 1A1 and 1A2 show an example of an image of a herd of pigs collected in a cage / gondola, that represents a high stocking density (Fig. 1A1), and an image with segmented frames isolating only pixels containing pigs (Fig. 1A2);

[0027] Figs. 1B1 and 1B2 show a graphical representation of the movement of a herd of pigs, wherein Fig. 1B2 shows a vector plot of the pigs depicted in Fig. 1B1; and

[0028] Fig. 2 shows the average estimated motion in each frame of a video of a stunning process as a function of time. Time point 0 seconds is when the gondola starts descending and the CO2 concentration increases.

[0029] DETAILED DISCLOSURE OF THE INVENTION

[0030] The system of the invention

[0031] In its first aspect, the invention provides a system for measuring the behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning method in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method.

[0032] The system may be characterized by comprising the following essential elements: a vision means (2), in communication with a processing means (3), which vision means (2) is configured to monitor, and to obtain video data, on the entire group of animals (1) subjected to the stunning process; and the processing means (3), in communication with the vision means (2), which processing means (3) is configured for collecting and computing data obtained from the visions means (2); in which system computing by the processing means (3) is accomplished using an artificial intelligence (Al) algorithm, to identify and sum up the activity of the animals, as observed by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals is compromised.

[0033] In one embodiment of the system of the invention, the animals are pigs, intended for slaughter at a slaughterhouse.

[0034] In another embodiment, the stunning process is a gas stunning process, that is accomplished in a CCh-enriched atmosphere.

[0035] In a further embodiment, the animals are placed in a gondola (4) intended for immersion into the CCh-enriched atmosphere.

[0036] In the context of this application, a CCh-enriched atmosphere indicates an atmosphere of a concentration of about 80% or more of CO2.

[0037] In an even further embodiment, the behavioral activity and changes in activity of a group of animals is estimated by an optical flow estimation algorithm / model.

[0038] The stunning process

[0039] The system and method of this invention relate to gas stunning of animals in groups, and the stunning process may be defined as the time from the animals (1) enter the stunning chamber / gondola / cage / fold / container / box (4) and are subjected to a gas stunning process, and until the animals leave the stunning chamber after a concluded stunning process.

[0040] Gas stunning may be accomplished using a gas or a mixture of gases, e.g., inert gasses like argon, nitrogen and dinitrogen oxide, or similar gasses allowed for stunning processes.

[0041] In one embodiment, the stunning process is a gas stunning process, that is accomplished in a CCh-enriched atmosphere of about 80% CO2 or more.

[0042] The vision means

[0043] For measuring the behavior of the animals (1), collected in the gondola / stunning chamber / cage / fold (4), the system of the invention shall comprise a vision means (2), in communication with the processing means (3), which vision means (2) is configured to monitor the animals during the stunning process.

[0044] In the context of this application, a vision means represents a combination of hardware and software capable of providing operational guidance to other devices based on the capture and processing of images, and usually rely on digital sensors protected inside an imaging device, especially an industrial camera, with specialised optics to acquire images, so that computer hardware and software can process, analyse, and measure various characteristics for decision making. A vision system typically comprises individual elements like lighting, one or more cameras with a lens, an image sensor, a vision processing means, and communication means. The lens captures the image and presents it to the sensor in the form of light. The sensor in a machine vision camera converts this light into a digital image, which is then sent to the processor for analysis. Lighting illuminates the part of the image to be inspected and allows its features to stand out so they can be clearly seen by the camera.

[0045] The cameras for use according to the invention may be any suitable camera / area- scan camera, e.g., a monochrome camera, a colour camera (RGB), multispectral camera (e.g., including both visible and infrared wavelengths), or a 3C / depth camera (e.g., a time-of-flight camera or a stereo camera system), or a combination hereof.

[0046] The vision means (2) for use according to the invention shall be in communication with the processing means (3).

[0047] Each camera used according to the invention shall be configured for recording one or more images, and for communicating the recorded data to the processing means (3).

[0048] The processing means

[0049] For collecting and computing data obtained from the individual elements of the vision means (2) used in the invention, the system of the invention shall comprise one or more, optionally inter-related or inter-connected, processing means (3). If two or more processing means are employed, these processors may be in inter-communication with one or more of the other processing means.

[0050] The processing means (3) for use according to the invention may be any available computation device, processor, controller, GPU, CPU, PLC, and / or PC.

[0051] For performing the necessary processing actions, the processing means (3) for use according to the invention shall be in communication with, and / or configured for processing, incl. computing, controlling, analysing, and / or calculating, digitalized data obtained by the vision means (2).

[0052] The method of the invention

[0053] In another aspect, the invention provides a method for measuring the behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning method in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method.

[0054] The method of the invention may be characterized by comprising the following subsequent steps: (i) subjecting the group of animals (1) to be subjected to the stunning process, to monitoring by a vision means (2), configured to monitor all animals during the stunning process; and

[0055] (ii) processing the data, obtained according to step ii, in a processing means (3); in which method the processing means (3) is configured for running an artificial intelligence (Al) algorithm, and which processing means (3) is configured to identify and sum up the activity of the animals (1), as monitored by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals are compromised.

[0056] In one embodiment of the method of the invention, the stunning process is a gas stunning process.

[0057] In another embodiment of the method of the invention, the animals are pigs, intended for slaughter at a slaughterhouse.

[0058] In another embodiment, the animals are placed in a gondola (4) intended for immersion into the CCh-enriched atmosphere.

[0059] In the context of this application a CCh-enriched atmosphere indicates an atmosphere of a concentration of about 80% or more CO2.

[0060] In a further embodiment, the pigs are placed in a gondola / container / box (4) and are exposed to a gas, or a gas mixture, for the purpose of stunning or stunning to kill.

[0061] In an even further embodiment, the behavioral activity, and changes in activity of the group of animals, is estimated by an optical flow estimation algorithm / model.

[0062] EXAMPLE

[0063] The invention is further illustrated with reference to the following example, which are not intended to be in any way limiting to the scope of the invention as claimed.

[0064] Example

[0065] In this example, we report the behavior of a group of pigs placed in a gondola, that is subsequently immersed in a CCh-enriched atmosphere. During this process, the group of pigs is monitored by a vision means (i.e., a video camera) covering all relevant parts of the gondola, and configured for recording images that were communicated to, and processed in, the processing means.

[0066] The level of activity per time unit is determined using an artificial intelligence model in order to evaluate the relative change in the pigs' position between video frames, and the activity level is calculated from segmented frames isolating only pixels containing pigs. Frame to frame activity is predicted with a deep learning model, with a Recurrent All-Pairs Field Transform (RAFT) architecture, trained on the KITTI dataset, essentially as described by Teed and Deng, RAFT: Recurrent All-Pairs Field Transforms for Optical Flow, Princeton University, 25 August 2020 (see, e.g.,

[0067] This RAFT model takes two frames at a time and calculates the optical flow, i.e., the velocity of movement in each pixel, by comparing the two frames. By feeding the model with data from the current and the previous frame, we get velocity data for all pixels in each frame.

[0068] For each frame we calculate the average speed of all pixels containing pigs. We then have a measurement of the level of activity for each frame. The level of activity is measured as the level activity per frame in each gondola.

[0069] The amount of activity in each gondola can be calculated based on activity measurements in all frames. This can, e.g., be done in the period from the pig's first response to the gas, and till the time the last pig loses balance and is not able to stand (i.e., loss of posture (LOP)). Or it can be done more generally if for example a slaughterhouse wants to monitor and document that the stunning proceeds calmly and that animals react to a limited extent to the gas that is supplied. A threshold value can be set and used in real time, to alarm if the animals react too strongly to the gas and interventions are needed.

[0070] DETAILED DESCRIPTION OF THE DRAWINGS

[0071] The present invention is further illustrated by reference to the accompanying drawing, in which:

[0072] The present invention is further illustrated by reference to the accompanying drawing, in which:

[0073] Figs. 1A1 and 1A2 show an example of an image of a herd of pigs collected in a cage / gondola, that represents a high stocking density (Fig. 1A1), and an image with segmented frames isolating only pixels containing pigs (Fig. 1A2), and the figures illustrate that when the pigs stand close together there will be parts of the pigs’ body that are not visible at all given times, and this makes it difficult to make manual records of their behavior and an image with segmented frames isolating only pixels containing pigs;

[0074] Fig. 1B1 and 1B2 show a graphical representation of the movement of a herd of pigs, wherein Fig. 1B2 shows a vector plot of the pigs depicted in Fig. 1B1, and in which the movements are shown as vector plots, and the vectors indicate the speed and direction of the movement at the given point of the vector. It is the average speed of movement, i.e., the average length of all these vectors, that can be used to measure the activity of the pigs in the stunner. This means that if the pigs move a lot and have many rapid movements, we get a result in high activity in the stunner; and

[0075] Fig. 2 shows the average estimated motion in each frame of a video of a stunning process as a function of time. Time point 0 seconds is when the gondola starts descending and the CO2 concentration increases.

Claims

CLAIMS1. A system for measuring the behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning method in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method, which system comprises the following essential elements: a vision means (2), in communication with a processing means (3), which vision means (2) is configured to monitor, and to obtain video data, on the entire group of animals (1) subjected to the stunning process; and the processing means (3), in communication with the vision means (2), which processing means (3) is configured for collecting and computing data obtained from the visions means (2); in which system computing by the processing means (3) is accomplished using an artificial intelligence (Al) algorithm, to identify and sum up the activity of the animals, as observed by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals is compromised.

2. The system according to claim 1, wherein the group of animals are pigs, intended for slaughter at a slaughterhouse.

3. The system according to claim 1, wherein the stunning process is a gas stunning method.

4. The system according to claim 1, wherein the animals are placed in a gondola / container / box (4) and are exposed to a gas, or a gas mixture, for the purpose of stunning or stunning to kill.

5. The system according to claim 1, wherein the behavioral activity is estimated by an optical flow estimation algorithm / model.

6. A method for measuring behavioral activity and changes in activity of a group of animals (1), when subjected to a stunning process in a testing or research situation, in a slaughterhouse, or in a stable system where it is permitted to stun and euthanize animals in groups with a gas method, which system comprises the following essential elements:(i) subjecting the group of animals (1) to be subjected to the stunning process, to monitoring by a vision means (2), configured to monitor all animals during the stunning process; and(ii) processing the data, obtained by the vision means (2) according to step ii, in a processing means (3); in which method the processing means (3) is configured for running an artificial intelligence (Al) algorithm, and which processing means (3) is configured to identify and sum up the activity of the animals (1), as monitored by the vision means (2), which behavioral activity is used as an indication of the way the animals respond to the given condition and to evaluate if the welfare of the animals are compromised.

7. The method of claim 6, wherein the stunning process is a gas stunning process.

8. The method of claim 6, wherein the group of animals are pigs, intended for slaughter at a slaughterhouse.

9. The method of claim 8, wherein the pigs are placed in a gondola / container / box (4) and are exposed to a gas, or a gas mixture, for the purpose of stunning or stunning to kill.

10. The method of claim 6, wherein the behavioral activity is estimated by an optical flow estimation algorithm / model.

Images