Technical Field

[0001] The present disclosure relates generally to a method and a system for monitoring a horse to predict foaling.

Background

[0002] Horse breeding techniques have been developed by humans since ancient times. One of the most vital steps is when a mare gives birth to her foal. Even though the foaling process occurs independently of human interventions, sometimes this intervention is essential in order to avoid serious injuries and/or death of the horse or the foal.

[0003] Horse monitoring is typically done by owners or staff of a breeding facility, zoos, farms or any place that has horses. This monitoring is usually labor intensive and therefore expensive, since it is not possible to predict with precision when a foaling process will happen. Furthermore, the animals cannot be monitored 24 hours a day by staff or owners.

[0004] In cases when a person detects a foaling process taking place and some issue is identified, it is usually too late for contacting and obtaining a professional help that would arrive on time for a veterinary intervention or assistance.

[0005] Some solutions in the art include one or more sensors connected to the horse body for continuously monitoring horse parameters, such as heart rate, laid down/up positions etc. However, those solutions may detect a foaling process that is taking place at the moment and thus are not able to predict and alert a staff/owner in advance about the foaling process.

[0006] Therefore, there is a need of a monitoring process of a horse that is able to avoid or at least reduce the above-mentioned problems. Summary

[0007] It is a first aspect of this disclosure to present a computer-implemented method for monitoring a horse to predict foaling and create an alarm that mitigates, alleviates or eliminates one or more of the above-identified deficiencies in the art singly or in combination. Another aspect of this disclosure is to present a system for monitoring a horse, predict foaling and create an alarm.

[0008] The first aspect is solved by providing a computer-implemented method for monitoring a horse to predict foaling and create an alarm, comprising the steps of monitoring a horse in a confined area by recording image data from an image recording device, identifying when the horse enters the confined area, starting to monitor horse movement levels for subsequent, equal length, time periods when the horse enters the confined area, comparing the movement levels with previously recorder movement levels. Further, an alarm is generated if a threshold is exceeded, the threshold being any one of: the movement level of a first time period after the horse entered the confined area exceeds a first threshold for an average first time period movement level, the movement level of a number of subsequent time periods exceed a second threshold for an average movement level of the same subsequent time periods from the horse entered the confined area.

[0009] One exemplary effect of this computer-implemented method is that it is possible to predict foaling and create an alarm by using an image recording device only. Thus, there is no need of using extra devices, such as devices attached to the horse in order to know its location or position. Furthermore, the method according to the disclosure advantageously predict a foaling event in an accurate manner by comparing actual movement level with previously recorded movement level, using an image recording device. The predicted foaling event may then trigger an alarm so one or more subjects may be notified and act according to the alarm received. In other words, one or more subjects may be notified before the foaling process starts to happen. [00010] A previous movement may be recorded by an image recording device such as a camera. Previous movement records of a horse that enters a confined area may provide a standard behavior pattern for that specific horse. Hence, the standard behavior pattern may then be compared to current behavior pattern determined by the current movement record of the horse. In case the current movement record exceeds a first threshold, an alarm may be triggered.

[00011 ] Alternatively, an average movement level for a number of subsequent time periods may be compared to the current movement level of the same subsequent time periods from the time since the horse entered the confined area. In case the current movement level of the same subsequent time periods exceeds a second threshold, an alarm may be triggered. Both first and second thresholds independently provide an accurate horse foal prediction.

[00012] According to one exemplary embodiment the equal length time periods are time periods of for example any one of 1 , 5, 10, 15, 20, 30, 60 or 120 minutes. In another example the equal length time periods are for each horse determined by the horse movement level. Such determination could for example be conducted via an artificial intelligence engine or average calculations.

[00013] According to one exemplary embodiment, the first time period is delayed at least 30 minutes after the horse enters the confined area. The first time period may be delayed at least 45 minutes, 60 minutes or 90 minutes after the horse enters the confined area. One exemplary effect of this embodiment is that it was noted that an even higher foaling prediction accuracy and/or a reduced number of false alarms may be obtained when the first period is delayed by at least 30 minutes. This effect may happen since the horse may naturally move substantially more when it has just entered the confined area, for instance for recognizing the area or searching the limits of the confined area. Furthermore, the horse movement level during the first 30 minutes in the confined area is more affected to what happened before the horse entered that confined area, e.g., if the horse ate, ran or trained [00014] According to another exemplary embodiment, the average movement levels are any one, or a combination, of movement levels recorder from multiple days for a specific horse, movement levels recorder from multiple days for a group of horses, and movement levels recorder for a general population of horses. One exemplary effect of this embodiment is that the horse movement may not only be compared to its own average movement, but also to average movements of a group of horses and/or a general population of horses. The comparison of movement levels recorder from multiple days for the specific horse and/or a group of horses and/or the general population of horses may be used to predict a foaling event for the specific horse. Further, movement data from a group of horse and/or movement data from the general population of horses may be used in case there are missing data from the specific horse. In this case, a reliable foaling prediction may also be obtained.

[00015] According to one exemplary embodiment, the horse movement level comprises information about for example if the horse is laying down, standing up, standing still, or moving together with the information how much the horse is moving. It is thus one advantage with the horse movement level as described herein and as monitored by the image recording device that the horse movement level provides relevant data for analysis independently if the horse is for example standing or lying down.

[00016] According to one exemplary embodiment, the average movement levels comprise only the last time period of a defined time period. The last time period may be substantially shorter than the defined time period. A defined time period may be selected from 12 hours, 1 day, 2 days, 4 days and 7 days. If for instance a defined time period is one day, the average movement level considers only the last time period of the day.

[00017] According to one exemplary embodiment, the average movement level of multiple time periods are averages of aggregated values of measurements within the time periods and averages of the aggregated time periods. Aggregated values may be measurements obtained approximately every second. The aggregated values may than be averaged per minute in order to obtain an average minute value. Further, the average minute levels may be averaged in for instance every 5 minutes, 10 minutes, 15 minutes, 20 minutes, 30 minutes and combinations thereof.

[00018] One advantageous effect is that substantial computational processing is avoided by the averaging the aggregated values. The computational processing savings generated by compressing data promote a more effective calculation. Further, less storage space is demanded.

[00019] According to another exemplary embodiment, the at least one of the first and second thresholds is dependent of the time of day. The time of the day may be in the morning, in the afternoon or in the evening. One exemplary effect is that depending on the time of the day, horse movement and/or behavior may change, so defining a first and second threshold accordingly may increase the predictability of the method.

[00020] According to another exemplary embodiment, the time periods has a length between 5 and 60 minutes. The time period may have a length of 5, 10, 20, 30, 40, 50 or 60 minutes. One exemplary effect of this embodiment is that data from 5 to 60 minutes may provide an ideal range for data collection, thereby improving the foaling prediction and reducing or avoiding false alarms.

[00021] According to another exemplary embodiment, missing data in the average movement levels is replaced with average movement data from the current time period. One exemplary effect of this embodiment is that the computer implemented method may still provide a horse monitoring that predicts foaling and create an alarm. The method still works even if data is missing by replacing the missing data with average movement data from the current time period.

[00022] According to another exemplary embodiment, the horse is identified in the image data from the image recording device by image recognition. The image recording device may be any device suitable for image recognition, such as a camera, video camera, radar, a lidar, an infrared camera and combinations thereof. [00023] According to another exemplary embodiment, the horse movement levels are calculated from the image data by at least the steps: masking an area around the horse to define pixels creating a mask relating to said horse, and calculating pixel movement of the mask.

[00024] According to another exemplary embodiment, the horse movement levels are calculated from the image data by at least the steps: masking an area around the horse to define an area in the image relating to said horse, and calculating the difference in the mask values between two or several images with masks.

[00025] According to another exemplary embodiment, generating the alarm comprise the steps: determine if another alarm has been generated within a threshold time, generate an alarm to a user if no alarm been generated within the threshold time and cancel subsequent alarms generated until expiry of the threshold time. One exemplary effect of this embodiment is to avoid generating multiple alarms to a user due to the same event. After the alarm is first triggered due to a threshold being exceeded, subsequent alarms will be canceled until the threshold time is expired.

[00026] According to another exemplary embodiment, generating the alarm comprise the step: transmit an alert to a personal telecommunication device. A personal telecommunication device may be any device suitable for telecommunication. A non-exhaustive list of personal telecommunication device comprises a phone, a mobile phone, a personal computer, a notebook, a tablet, a wearable device, a smartwatch, a smart band and combinations thereof.

[00027] A second aspect of the disclosure concerns a system for monitoring a horse, predict foaling and create an alarm, wherein the system comprises an image recording device, a memory, a central processing unit, and alarm means. Further, the image recording device is adapted to be arranged to monitor a horse in a confined area, the system is adapted to by means of image data from the image recording device identify when the horse enters the confined area, start monitoring horse movement levels for subsequent, equal length, time periods when the horse enter the confined area. Further, the system compares the movement levels with previously recorder movement levels and activate the alarm means if a threshold is exceed, wherein the threshold is any one of: the movement level of a first time period after the horse entered the confined area exceeds a first threshold for an average first time period movement level, the movement level of a number of subsequent time periods exceed a second threshold for an average movement level of the same subsequent time periods from the horse entered the confined area.

[00028] One exemplary effect of the system is that a horse foaling can be predicted, and an alarm can be triggered before the foaling process begins. Therefore, the system can predict a foaling process and trigger an alarm without the need of human monitoring, e.g. staff members or the owner.

[00029] According to one exemplary embodiment, the image recording device is any one of a camera, a video camera, and an infrared camera. The image recording device of the system may be any one of a camera, video camera, radar, lidar, infrared camera or combinations thereof.

[00030] According to one exemplary embodiment the system may further comprise an artificial intelligence (Al) engine for any one of the steps, for example setting the length of equal length time periods, identifying the horse, thresholds, or identifying a horse movement level depending on the horse is standing, laying down, standing still, or moving around.

[00031] In one embodiment, the artificial intelligence engine uses machine learning and neural networks to improve at least one of setting the length of equal length time periods, identifying the horse, thresholds, or identifying a horse movement level depending on the horse is standing, laying down, standing still, or moving around.

[00032] According to another exemplary embodiment, the system performs the method according to any one of the previous embodiments.

[00033] According to another aspect the system may monitor abnormalities and abnormal motion patters to detect and alarm if the horse behaves out of the normal. This is for example advantageous for detecting other triggers than foaling, such as colic, diseases, or any problem occurring that affects the horse. The image recording device can further be used to provide live or accumulated feeds of image and/or video to a user via for example the same device as the user receives an alarm via.

[00034] Other parameters such as the length of rest, amount of movement, numbers of times the horse gets up standing and/or down in a time period could further be used to understand and analyze both foaling and other triggers.

Brief description of the drawings

[00035] Fig. 1 illustrates a flow chart describing a computer implemented method for monitoring a horse, according to one embodiment.

[00036] Fig. 2 illustrates a flow chart describing a computer implemented method for monitoring a horse having two triggers, according to one embodiment.

[00037] Fig. 3 illustrates a blocking diagram of a system for monitoring a horse, according to one embodiment.

Detailed description

[00038] Figure 1 shows a computer-implemented method 10 for monitoring a horse to predict foaling and create an alarm. The method comprises a step 100 of monitoring a horse in a confined area, and the monitoring step 100 may use an image recording device. The image recording device is positioned in a place where it can record the whole confined area where the horse may move. Some confined areas may be recorded by more than one image recording device so all area can be accurately monitored. Furthermore, the image recording device may be stationary or may comprise means for moving in case the image recording device needs to be repositioned.

[00039] In step 110, it is identified when the horse entered a confined area. The step 110 is a relevant step since on this step it is detected when the horse enters the confined area and a time period can be counted from this moment. The horse may be identified by the image recording device using any technique known in the art, for instance by masking an area around the horse to define pixels and creating a mask related to the horse or masking an area in the image relating to said horse. This step 110 may usually be used to identify one horse per confined area, but it may be applied to monitor two or more horses in the same confined area.

[00040] In step 120, horse movement levels are constantly monitored and stored. Horse movement levels may be monitored by any technique known in the art, for instance by by masking an area around the horse to define pixels, creating a mask related to the horse, and calculating pixel movement of the mask. The horse movement levels are monitored in a subsequent, equal length, time periods manner, so that this data may be compared to future movement level data using the same parameters (e.g. equal length, time period etc.). The time period may be 15 minutes since the horse entered the confined area. The time period may be 60 minutes since the horse entered the confined area. The time period may be 180 minutes since the horse entered the confined area. Alternatively, the time period may be the morning, the afternoon or the evening.

[00041] Movement levels in step 120 may be calculated in batches of 15 minutes, therefore one hour is measured as 4 batches of 15 minutes each.

[00042] In step 130, the current movement level of a horse is compared to previous movement level recorded by the image recording device. The movement level of a horse is compared to a previous movement level on an equivalent time period. For instance, movement level 15 minutes after entering the confined area can be compared to previous 15 minutes after entering the confined area. The movement level 15 minutes after entering the confined area can be compared to an average of previous 15 minutes after entering the confined area. The average may be the average of the last 7, 8, 9, 10, 11 or more days.

[00043] The average calculation may include only the last event of a time period such as a day. For instance, in case a horse enters a confined area at 2pm, leaves at 4.30pm, returns to the confined area at 8pm and leaves the next day at 8am, the average calculation considers only the last event, i.e. , the average movement levels from 8pm to 8am.

[00044] In step 140, it is measured whether a threshold was exceeded. The threshold may be 130, 150, 180, 200 or 300% higher current movement level compared to the averaged movement levels. In case the threshold is not reached, step 140 ends and the method return to step 120, i.e. monitoring the horse movement levels.

[00045] In step 150, an alarm is triggered when the current movement level reaches a threshold when compared to previous averaged movement levels. The alarm may be triggered every 15 minutes if any of the threshold values are exceeded. The alarm may be any means for inducing a subject to perceive that the alarm was triggered. The alarm may be a sound, a beep, a light, a vibration and combinations thereof. The alarm may be sent to one or more personal telecommunication devices, such as smartphones, computers, or other personal communication devices.

[00046] When the alarm is triggered by any threshold, all other alarms may be paused for the next 240, 260, 285, 300, 325 minutes.

[00047] The alarm may not be sent during the first 15, 30, 45, 60, 90, 120 minutes since the horse has entered the confined area. Alternatively, the alarm may not be sent after 90, 120, 180 minutes have passed.

[00048] Fig.2 shows a computer-implemented method 20 where two different thresholds may be used as an alarm trigger. The steps of monitoring a horse 200, identifying when the horse enters a confined area 210, monitoring horse movement levels 220 and comparing movement levels with previous movement levels 230 are as previously described.

[00049] In step 240, it is measured whether a first threshold was exceeded. The first threshold may be exceeded when the movement level of a first time period after the horse entered the confined area is higher than the average movement level of previous first time periods. Alternatively, the first threshold may be exceeded when the movement level of a number of subsequent time periods after the horse entered the confined area is higher than the average movement level of the same subsequent time periods.

[00050] In case the first threshold 240 is exceeded, a step 250 of triggering the alarm is followed.

[00051 ] In case any of the thresholds are not exceeded, it is checked in the next step 260 whether the second threshold was exceeded. The second threshold may be exceeded when the movement level of a first time period after the horse entered the confined area is higher than the average movement level of previous first time periods. Alternatively, the second threshold may be exceeded when the movement level of a number of subsequent time periods after the horse entered the confined area is higher than the average movement level of the same subsequent time periods.

[00052] In case the first threshold is related to the movement level of a first time period, the second threshold will be the movement level of a number of subsequent time periods.

[00053] In case the first threshold is related to the movement level of a number of subsequent time periods, the second threshold will be the movement level of a first time period.

[00054] In case the second threshold 260 is exceeded, a step 270 of triggering the alarm is followed.

[00055] When both threshold steps 240 and 260 do not lead to triggering the alarm, the method return to the step 220 of monitoring the horse movement levels.

[00056] Fig. 3 depicts an illustrative block diagram of a system 30 for monitoring a horse in which a set of instructions for causing the system to perform any one of the methods discussed herein may be executed. The system 30 may comprise 1 , 2, 3, 4, 5, 6, 10, 20 or more devices. [00057] The system 30 for monitoring a horse may include at least one central processing unit 320, a memory 310, a network adapter 330, an image recording device 340, and alarm means 350, and storage means 360 that are in connection to a bus 300. The bus 300 represents one or more separate buses connected by suitable controllers, adapters and/or bridges, and may be any form of network and/or internal bus, alone or in combination.

[00058] The central processing unit 320 may be one or more processor devices known in the art, such as microprocessors, and the central processing unit 320 is configured to execute instructions related to the steps of the method as described.

[00059] The image recording device 340 may be any device suitable for recording an image, such as camera, video camera, radar, a lidar, an infrared camera and combinations thereof.

[00060] The alarm means 350 may be any means for inducing a subject to perceive that the alarm was triggered. The alarm means 350 may be a sound, a beep, a light, a vibration and combinations thereof. The alarm means 350 may be sent and triggered in one or more personal telecommunication devices, such as smartphones.

[00061] The storage means 360 is a computer-readable medium on which data related to horse movement levels are stored.

[00062] The system 30 may be several independent system devices 30;30a,30b;30a...30n. The system devices 30;30a,30b;30a...30n are in one embodiment connected via a network and each comprise one or multiple of the devices 310, 320, 330, 340, 350, 360 as illustrated in Fig. 3. The system 30 can monitor a horse, predict foaling and create an alarm, wherein the alarm is triggered by at least one threshold as previously described.