Weak chicken detection device and weak chicken detection program

Abstract

The present invention detects a weak chicken from a decrease in the activity level of a chicken, and comprises: an imaging device 2 that images a plurality of chickens housed in a poultry facility; and a weak chicken detection unit 33 that detects a weak chicken or identifies an area in which the weak chicken is included, on the basis of images captured by the imaging device 2.

Description

Weak chicken detection device and weak chicken detection program

[0001] The present invention relates to a device for detecting weak chickens and a program for detecting weak chickens.

[0002] In poultry farming, such as that of laying hens, outbreaks of infectious diseases with high mortality rates have become a serious problem worldwide, and the need for disease prevention measures is urgently needed. One method for doing so is to detect and respond to chickens with reduced activity levels early. Along with detecting dead chickens, early detection of weakened chickens with reduced activity levels is also desirable, but this currently relies on visual inspection by workers on patrol, which is an extremely difficult task in today's high-density, large-scale poultry farming.

[0003] In recent years, in order to automate the detection of dead chickens, dead chicken detection devices have been proposed that detect dead chickens by analyzing images captured by an imaging device, as shown in Patent Document 1, for example. Such dead chicken detection devices have been achieving some success in automating the detection of dead chickens.

[0004] However, detection of weakened chickens with reduced activity levels still relies on visual inspection by workers patrolling, which is extremely difficult and insufficient in high-density rearing environments. Therefore, currently, weakened chickens are discovered only after they have died. It is known that infected chickens develop symptoms after an incubation period (latent infection period), and that chickens spread the virus to their surroundings between the time they develop symptoms and the time they die. This is why early detection and early response are so important. Note that in Patent Document 1, depending on how the threshold for detecting dead chickens is set, weakened chickens in addition to dead chickens may merely be identified as dead chickens as noise, and weakened chickens are not actively detected.

[0005] JP 2024-39126 A

[0006] Therefore, an object of the present invention is to detect weak chickens based on a decrease in their activity level.

[0007] In other words, the weak chicken detection device of the present invention is characterized by comprising an imaging device that images multiple chickens housed in a poultry farming facility, and a weak chicken detection unit that detects weak chickens or identifies areas containing weak chickens based on images captured by the imaging device.

[0008] This weak chicken detection device can detect weak chickens or identify areas containing weak chickens by focusing on their health status, such as decreased activity levels (depression), based on images captured by the imaging device. As a result, it is possible to quickly detect early symptoms of illness or decreased activity levels due to deterioration of the air, water, feed, or other environmental factors caused by problems with equipment (such as drinking fountains, feeders, or ventilation fans), and take early action to minimize damage.

[0009] It is desirable that the weak chicken detection unit detects weak chickens or identifies the area containing the weak chickens based on the images captured by the imaging device and the elapsed time calculated from the time of capturing the images. With this configuration, the degree of decline in chicken activity over time can be determined, enabling weak chickens to be detected with high accuracy.

[0010] Specifically, it is desirable for the weak chicken detection unit to classify weak chickens into multiple attributes based on the elapsed time. For example, weak chickens can be classified into multiple attributes based on the elapsed time until they reach state A (e.g., a crouching state). If the elapsed time until state A is T1 (e.g., within 5 hours), they can be classified into a first attribute. If the elapsed time until state A is T2 (e.g., 5 hours or more but within 24 hours), they can be classified into a second attribute. If the elapsed time until state A is T3 (e.g., 24 hours or more but within 3 days), they can be classified into a third attribute.

[0011] The weak chicken detection unit preferably estimates the cause of the weak chicken's weakness based on the elapsed time and classifies the weak chicken into multiple causes of weakness as the multiple attributes. Here, the multiple causes of weakness (multiple attributes) can include various infectious diseases, respiratory symptoms, gastroenteritis, neurological symptoms, heatstroke, or factors related to the rearing environment (e.g., overcrowding, lack of sunlight, insufficient ventilation, hyperventilation, lack of water, lack of food, etc.).

[0012] The weak chicken detection unit preferably estimates the degree of weakness of the weak chickens or the seriousness of the cause. With this configuration, appropriate measures can be taken depending on the degree of weakness or the seriousness of the cause.

[0013] When weak chickens are detected based on images, dead chickens may be mistakenly detected as weak chickens. For this reason, it is desirable that the imaging device captures visible light images or infrared images and thermal images, and that the weak chicken detection unit provisionally detect weak chickens or dead chickens based on the visible light images or infrared images, and determine whether the provisionally detected weak chickens or dead chickens are weak chickens or dead chickens based on the thermal images. With this configuration, if body temperature can be detected by the thermal images, the chickens can be determined to be weak chickens, and if body temperature cannot be detected by the thermal images, the chickens can be determined to be dead chickens.

[0014] A specific embodiment of the weak chicken detection unit is to detect weak chickens using a learning model obtained by machine learning using multiple images of weak chickens as training data, or a learning model obtained by machine learning using multiple images of weak chickens and the times at which they were taken as training data. By using a learning model in this way, weak chickens can be detected with high accuracy.

[0015] The weak chicken detection device of the present invention preferably further includes an urgency calculation unit that calculates the urgency of responding to weak chickens based on the detection results of the weak chicken detection unit. With this configuration, it is possible to notify the operator of the need to respond to detected weak chickens, such as the need to respond immediately or to respond after observing the situation for a certain period of time, depending on the condition of the detected weak chickens.

[0016] In addition, the weak chicken detection program of the present invention is a weak chicken detection program that detects weak chickens, which are weak chickens, in multiple compartments that house multiple chickens, and is characterized in that it has a computer that functions as a weak chicken detection unit that detects weak chickens in each compartment based on images of each compartment taken at multiple points in time by an imaging device and the times at which those images were taken.

[0017] According to the present invention configured in this way, weak chickens can be detected by focusing on the health status of the chickens in each section, such as a decrease in activity level over time (depression).

[0018] Fig. 1 is a plan view schematically showing an egg-laying hen house system according to one embodiment of the present invention; Fig. 2 is a functional configuration diagram of a detection and calculation device according to the embodiment; Fig. 3 is an image taken of the state on the night of the second day after inoculation in an observation experiment; Fig. 4 is an image taken of the state on the morning and night of the third day after inoculation in an observation experiment; Fig. 5 is an image taken of the state on the night of the fourth day after inoculation in an observation experiment.

[0019] An embodiment of an egg-laying chicken house system equipped with a weak chicken detection device according to the present invention will be described below with reference to the drawings. Note that, for ease of understanding, all of the drawings shown below are drawn in a schematic manner, with appropriate omissions or exaggerations. Identical components are designated by the same reference numerals, and their descriptions will be omitted where appropriate.

[0020] <1. System Configuration> As shown in FIG. 1 , the egg-laying hen house system 100 of this embodiment includes an egg-laying hen house 10, which is a poultry farming facility having an artificial environment adjustment function such as an enclosed chicken house (windowless chicken house) for egg-laying hens, and a weak chicken detection device 20 that detects weak chickens in the egg-laying hen house 10.

[0021] Here, weak chickens include chickens whose activity levels have decreased due to illnesses such as various infectious diseases, respiratory symptoms, gastroenteritis, neurological symptoms, and heatstroke, or chickens whose activity levels have decreased due to a poor rearing environment such as overcrowding, lack of sunlight, insufficient ventilation, hyperventilation, lack of water, or lack of food.

[0022] <2. Egg-laying hen house 10> The egg-laying hen house 10 is provided with a plurality of cage rows 12, each of which has a plurality of cages 11 for housing egg-laying hens arranged in one direction. In this embodiment, the cage rows 12 are arranged side by side from left to right as shown in Figures 1 and 2. The cage rows 12 are also arranged in three tiers, one above the other, but the number of tiers is not limited to this. In addition, the egg-laying hen house 10 is also provided with an egg collection belt arranged along the cage rows, drinking water dispensers provided in each cage, a feeding device that supplies feed to each cage, a ventilation fan that ventilates the egg-laying hen house 10, and air conditioning equipment.

[0023] <3. Weak Chicken Detector 20> The weak chicken detector 20 detects weak chickens in a plurality of compartments (here, a plurality of cages 11) housing a plurality of chickens.

[0024] Specifically, the weak chicken detection device 20 includes an imaging device 2 that captures images of multiple chickens housed in each cage 11, and a detection and calculation device 3 that detects weak chickens in each cage 11 based on the images of each cage 11 captured by the imaging device 2.

[0025] The imaging device 2 captures images of the chickens housed in each cage 11 and is movable by a moving device 4, such as a feeding device that moves along the cage row 12. In this embodiment, an imaging device 2 is provided for each cage row 12, but one imaging device 2 may be used for multiple cage rows 12. The imaging device 2 moves multiple times along the cage row 12, capturing images of the chickens housed in each cage 11 at multiple points in time. The imaging device 2 can also link time information indicating the time of capture to each captured image and output the images to the detection and calculation device 3.

[0026] The time when each cage 11 is imaged can be set appropriately, and may be a predetermined time of day, such as a predetermined time in the morning and / or a predetermined time in the evening, or may be every predetermined time interval such as every eight hours, or may be every predetermined number of days such as every day. Furthermore, the time when each cage 11 is imaged may be set to a time when the laying hens are most active, such as the peak of egg-laying, based on egg-laying information by time period according to the physiology of the laying hens.

[0027] The imaging device 2 may be one that captures still images or one that captures moving images, and may also be one that captures visible light images or one that captures infrared images.

[0028] The detection and calculation device 3 detects weak chickens in each cage 11 based on the image of each cage 11 captured by the imaging device 2. The detection and calculation device 3 is composed of a computer having a CPU, memory, input / output interface, AD converter, input device, display device, etc. The CPU and peripheral devices work together to perform each function of the detection and calculation device 3 based on a predetermined weak chicken detection program stored in the memory. The detection and calculation device 3 may be composed of a single computer or multiple computers.

[0029] Specifically, as shown in Figure 2, the detection and calculation equipment 3 has an image acquisition unit 31 that acquires an image from the imaging device 2, a recording unit 32 that records the image, and a weak chicken detection unit 33 that detects weak chickens based on the image or identifies the area in which the weak chickens are located.

[0030] Specifically, the captured image acquisition unit 31 acquires each captured image captured by the imaging device 2 together with time information indicating the time of capture and position information indicating the position of the image. Note that the position information in this embodiment is information that identifies the cage or its surroundings corresponding to each captured image, and the position information at the time when the imaging device 2 captured the image can be acquired from the moving device 4 that moves the imaging device 2. Furthermore, the captured position may be calculated from the travel time of the moving device 4 from a reference position, or the position information may be acquired by image recognition of an identifier such as a cage number included in the captured image.

[0031] The recording unit 32 records images of each cage 11 taken by the imaging device 2 at multiple points in time, in association with time information indicating the time at which each image was taken. Note that the time at which each cage 11 is taken will be shifted due to the travel time of the imaging device.

[0032] The weak chicken detection unit 33 detects weak chickens in each cage 11 based on images of each cage 11 taken at multiple points in time by the imaging device 2 and the times at which these images were taken. Specifically, the weak chicken detection unit 33 detects weak chickens in each cage 11 based on the elapsed time calculated from the times at which the images taken at multiple points in time were taken.

[0033] The weak chicken detection unit 33 detects weak chickens in the mth cage 11, for example, by comparing images taken at multiple points in time (e.g., t1, t2, t3, t4) in the mth cage 11.

[0034] For example, the weak chicken detection unit 33 may detect a chicken as weak if it determines that the activity level of the chicken has decreased in one or more of the images captured at multiple times t1 to t4. Here, the decrease in activity level can be determined using reference images that show each state of the chicken. Examples of reference images include images of a healthy chicken and images of a chicken in each stage of weakened state.

[0035] Furthermore, if the change in activity level between the image taken at time t1 and the image taken at time t4 is greater than the change in activity level between the image taken at time t1 and the image taken at time t2 or t3, it is possible to detect a weak chicken at time t4. Here, the change in the decline in activity level can be determined using reference images that show each state of the chicken. It is possible to digitize the above-mentioned reference image, digitize the state of the chicken in the image taken at each time, and calculate the difference between them to digitize the change in the decline in activity level.

[0036] In addition, it is desirable that the weak chicken detection unit 33 classifies weak chickens in each cage 11 into multiple attributes based on the elapsed time Δt (for example, t2-t1).

[0037] Specifically, the weak chicken detection unit 33 can classify weak chickens into multiple attributes based on the time Δt that has passed until the weak chicken reaches state A (e.g., a crouched state). If the time Δt that passed until state A was reached is T1 (e.g., within 5 hours), the chicken can be classified into the first attribute; if the time Δt that passed until state A was reached is T2 (e.g., 5 hours or more but within 24 hours), the chicken can be classified into the second attribute; and if the time Δt that passed until state A was reached is T3 (e.g., 24 hours or more but within 3 days), the chicken can be classified into the third attribute.

[0038] Furthermore, the weak chicken detection unit 33 can estimate the cause of weakness of weak chickens in each cage 11 based on the elapsed time Δt and classify the causes into multiple attributes. Here, the multiple causes of weakness (multiple attributes) can include various infectious diseases, respiratory symptoms, gastroenteritis, neurological symptoms, heat stroke, or factors related to the rearing environment (for example, overcrowding, lack of sunlight, insufficient ventilation, hyperventilation, lack of water, lack of food, etc.).

[0039] Specifically, the weak chicken detection unit 33 can determine and classify the cause of weakness as an acute infection if the elapsed time Δt until the chicken reached state A (e.g., a crouched state) is within a predetermined time, such as within 5 hours. Also, the weak chicken detection unit 33 can determine and classify the cause of weakness as a rearing environment factor (e.g., insufficient ventilation) if the elapsed time Δt until the chicken reached state B (e.g., a drooping head) is within a predetermined time range, such as 24 hours.

[0040] Furthermore, the weak chicken detection unit 33 detects weak chickens in each cage 11 using a learning model obtained by machine learning using multiple images obtained by photographing weak chickens at multiple points in time as training data, or a learning model obtained by machine learning using multiple images obtained by photographing weak chickens at multiple points in time and the times at which these images were photographed as training data. Here, the multiple images used as training data include images taken at multiple points in time of chickens weakened by each cause of weakness.

[0041] Additionally, the detection and calculation device 3 may have a detection notification unit 34 that notifies the detection of a weak chicken when the weak chicken detection unit 33 detects the weak chicken. This detection notification unit 34 may display information notifying the occurrence of an abnormality on a display device such as a display, or may use an indicator light or an alarm provided inside or outside the egg-laying chicken coop 10.

[0042] 4. Experimental Example Next, an observation experiment on weakened chickens that supports the present invention will be described.

[0043] The test virus (highly pathogenic influenza virus) was diluted to 10 ７ 10-fold serial dilutions were made up to 10４ ~10 ７ For each dilution step, 0.05 mL of diluted virus solution was intranasally inoculated into five chicks. Thereafter, the chicks were observed and their food and water were changed every day. They were then observed for 4 to 5 days.

[0044] Figure 3 shows the state of the chicks on the night of the second day after inoculation. The chicks in cages No. 5 to No. 7 were in good health, the chicks in cage No. 4 were listless, and the chicks in cage No. 3 were confirmed dead. The cage No. indicates the dilution factor multiplied by 10.

[0045] Figure 4(a) shows the state of the chicks on the morning of the third day after inoculation, and Figure 4(b) shows the state of the chicks on the night of the third day after inoculation. The chicks in cages No. 6 and No. 7 were in good health, while the chicks in cage No. 5 were listless in the morning, and by nightfall, two had died, and it was confirmed that the chicks in cage No. 4 had died.

[0046] Figure 5 shows the state of the chicks on the night of the fourth day after inoculation. The chicks in cages 6 and 7 were still in good health and were not infected with the test virus. It is believed that the test virus had disappeared due to dilution.

[0047] 5. Effects of this embodiment The weak chicken detection device 100 of this embodiment detects weak chickens in each cage 11 based on images of each cage 11 taken by the imaging device 2 at multiple points in time and the times at which those images were taken, making it possible to detect weak chickens by focusing on the health status of the chickens in each cage 11, such as a decrease in activity over time (depression). As a result, it is possible to quickly detect early symptoms of illness or a decrease in activity level due to a deterioration in the environment, such as air, water, or food, caused by problems with equipment (e.g., drinking water dispensers, feeders, ventilation fans, etc.), and to minimize damage through early response.

[0048] 6. Modified Embodiments of the Present Invention The present invention is not limited to the above-described embodiments.

[0049] In the above embodiment, weak chickens in each cage in a chicken coop are detected, but it is also possible to detect weak chickens in areas other than cages (for example, areas for free-range broilers, etc.).

[0050] The weak chicken detection unit 33 may also identify an area containing weak chickens based on the image captured by the imaging device 2 and location information indicating the image capture location. In this case, the weak chicken detection unit 33 may identify the cage 11 in which the weak chicken is housed, or the surrounding area including the cage 11.

[0051] Furthermore, the weak chicken detection unit 33 may estimate the degree of weakness of weak chickens or the seriousness of the cause. In this estimation, it is preferable to use a learning model obtained by machine learning using multiple captured images of weak chickens as training data.

[0052] The weak chicken detection unit 33 may also detect weak chickens from a single captured image rather than from images captured at multiple points in time. Here, the single captured image may be, for example, a single still image or a moving image of several seconds (e.g., 1 to 2 seconds). In this case, weak chickens can be detected by taking advantage of the chickens' tendency to become startled and become more active when the moving device 4 moves and approaches them.

[0053] For example, when the imaging device 2 attached to the moving device 4 approaches, a lively chicken will move around vigorously and stand up, and from the image captured at that time, it can be determined that the chicken is a lively chicken.

[0054] On the other hand, weak chickens tend to be slow to react or remain seated when the image capture device 2 attached to the moving device 4 approaches, so it is possible to determine that the chicken is a weak chicken from the captured image. In particular, this can be determined by analyzing a few seconds (for example, 1 to 2 seconds) of video.

[0055] Furthermore, there is a possibility that chickens detected as weak chickens are dead chickens. In order to be able to distinguish between weak chickens and dead chickens, the following configuration may be adopted.

[0056] Specifically, the imaging device 2 captures a visible light image or an infrared image and a thermal image. In this case, the imaging device 2 has an imaging element for capturing a visible light image or an infrared image and an imaging element for capturing a thermal image. The imaging device 2 may also generate a thermal image from a visible light image or an infrared image.

[0057] The weak chicken detection unit 33 provisionally detects weak or dead chickens based on the visible light image or infrared image, and determines whether the provisionally detected weak or dead chickens are weak or dead chickens based on the thermal image. Specifically, the weak chicken detection unit 33 can determine a chicken as weak if it can detect body temperature using the thermal image, and determines a chicken as dead if it cannot detect body temperature using the thermal image.

[0058] The weak chicken detection device 20 may further include an urgency calculation unit that calculates the urgency of responding to a weak chicken based on the detection results of the weak chicken detection unit 33. This urgency calculation unit calculates urgency information, such as whether immediate action is required or whether action should be taken after observing the situation for a certain period of time, depending on the condition of the detected weak chicken. Based on this urgency information, workers can take appropriate action on the detected weak chicken, the surrounding chickens, or the chicken house equipment.

[0059] Furthermore, the weak chicken detection unit 33 may be configured not only to detect weak chickens, but also to detect dead chickens and normal chickens, and classify the multiple chickens housed in the cage into three categories: dead chickens, weak chickens, and normal chickens.

[0060] Furthermore, the present invention is not limited to the above-described embodiment, and it goes without saying that various modifications are possible without departing from the spirit of the present invention.

[0061] According to the present invention, weak chickens can be detected from a decrease in the activity level of the chickens.

[0062] 100... Egg-laying chicken house system 20... Weak chicken detection device 2... Imaging device 33... Weak chicken detection unit

Claims

1. A weak chicken detection device comprising: an imaging device that captures images of multiple chickens housed in a poultry farming facility; and a weak chicken detection unit that detects weak chickens or identifies areas containing weak chickens based on the images captured by the imaging device.

2. A weak chicken detection device as described in claim 1, wherein the weak chicken detection unit detects weak chickens or identifies areas containing weak chickens based on images taken by an imaging device and the elapsed time calculated from the time at which the images were taken.

3. The weak chicken detection device according to claim 2, wherein the weak chicken detection unit classifies weak chickens into a plurality of attributes based on the elapsed time.

4. The weak chicken detection device according to claim 3, wherein the weak chicken detection unit estimates the cause of the weak chicken's weakness based on the elapsed time and classifies the cause into a plurality of weak chickens as the plurality of attributes.

5. A weak chicken detection device according to any one of claims 1 to 4, wherein the weak chicken detection unit estimates the degree of weakness of the weak chickens or the seriousness of the cause.

6. A weak chicken detection device as described in any one of claims 1 to 3, wherein the imaging device captures visible light images or infrared images and thermal images, and the weak chicken detection unit provisionally detects weak chickens or dead chickens based on the visible light images or the infrared images, and determines whether the provisionally detected weak chickens or dead chickens are weak chickens or dead chickens based on the thermal images.

7. A weak chicken detection device as described in any one of claims 1 to 3, wherein the weak chicken detection unit detects weak chickens using a learning model obtained by machine learning using multiple images obtained by photographing weak chickens as training data, or a learning model obtained by machine learning using multiple images obtained by photographing weak chickens and the times at which those images were photographed as training data.

8. A weak chicken detection device as claimed in any one of claims 1 to 3, further comprising an urgency calculation unit that calculates the urgency of dealing with weak chickens based on the detection results of the weak chicken detection unit.

9. A weak chicken detection program that detects weak chickens, which provides a computer with the function of a weak chicken detection unit that detects weak chickens or identifies areas containing weak chickens based on images captured by an imaging device.

Images