Automatic feeding method of bio-fermented feed

Abstract

The embodiment of the present application provides a kind of automatic feeding method of biological fermentation feed, comprising: obtaining the first monitoring image of feeding area and the second monitoring image of active area;Based on the first monitoring image, the biological fermentation feed in feeding area is carried out residual amount identification, and the biological fermentation feed residual amount is obtained;Based on the second monitoring image, the activity amount analysis of target object in active area is carried out, and the activity amount of target object is obtained;Based on the biological fermentation feed residual amount and the activity amount of target object, the feeding amount of biological fermentation feed is determined, and according to feeding amount, the feeding of biological fermentation feed in feeding area is carried out.The present application can identify the biological fermentation feed residual amount of feeding area, and the activity amount of target object in active area is analyzed through the second monitoring image, and the feeding amount of biological fermentation feed is determined according to the activity amount of target object and the biological fermentation feed residual amount of feeding area, which can more accurately feed the biological fermentation feed, improve the feeding accuracy of target object.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of feed feeding, in particular to an automatic feeding method of biological fermentation feed. BACKGROUND

[0002] In order to provide the meat quality of the farmed animals, the farmed animals are generally provided with activity space, so that the farmed animals can move in the activity space and are fed with biological fermentation feed. The feed made by fermenting the roughage with microorganisms is the biological fermentation feed. The biological fermentation feed is a biological fermentation feed which is made by using microorganisms and compound enzymes as biological feed fermentation agent strains to convert the feed raw materials into microorganism cell protein, bioactive small peptide amino acid, microorganism active probiotics and compound enzyme preparation. The roughage is rich in cellulose, hemicellulose, pectin substances and lignin, etc. However, the roughage is difficult to be directly digested and absorbed by the animals, and the animals will increase the intestinal burden and cause intestinal diseases after eating the roughage. The biological fermentation feed can not only make up the easily deficient amino acids in the conventional feed, but also rapidly convert the nutrients of other roughage raw materials to achieve the effect of enhancing the digestion, absorption and utilization. The traditional biological fermentation feed feeding is generally carried out in a timed and quantitative manner. However, since the movement of the farmed animals is uncertain, when the farmed animals move more, the biological fermentation feed is not enough to eat, and when the farmed animals move less, the biological fermentation feed is not eaten completely, which causes over-fermentation or waste. Therefore, the existing biological fermentation feed feeding has the problem of poor feeding accuracy. SUMMARY

[0003] The present application provides an automatic feeding method of biological fermentation feed, which aims to solve the problem of poor feeding accuracy of the existing biological fermentation feed feeding in a timed and quantitative manner. The residual amount of the biological fermentation feed in the feeding area is identified through the first monitoring image, and the activity amount of the target object in the activity area is analyzed through the second monitoring image. The feeding amount of the biological fermentation feed is determined according to the activity amount of the target object and the residual amount of the biological fermentation feed in the feeding area, so that the feeding of the biological fermentation feed can be more accurate, and the feeding accuracy of the target object can be improved.

[0004] In a first aspect, the present application provides an automatic feeding method of biological fermentation feed, which comprises the following steps:

[0005] obtaining a first monitoring image of a feeding area and a second monitoring image of an activity area;

[0006] based on the first monitoring image, identifying the residual amount of the biological fermentation feed in the feeding area to obtain the residual amount of the biological fermentation feed;

[0007] based on the second monitoring image, analyzing the activity amount of the target object in the activity area to obtain the activity amount of the target object.

[0008] determining a feeding amount of the bio-fermented feed based on the bio-fermented feed residual amount and the activity amount of the target object, and feeding the bio-fermented feed to the feeding area according to the feeding amount.

[0009] Optionally, the bio-fermented feed residual amount is obtained by identifying the bio-fermented feed residual amount in the feeding area based on the first monitoring image, and the method comprises the following steps.

[0010] performing first target detection on each frame of image in the first monitoring image to obtain a target image of the feeding area;

[0011] performing first image segmentation on the target image of the feeding area to obtain a background image of the feeding area and a bio-fermented feed image of the feeding area;

[0012] identifying the bio-fermented feed residual amount in the feeding area according to the background image of the feeding area and the bio-fermented feed image of the feeding area.

[0013] Optionally, the activity amount of the target object is obtained by analyzing the activity amount of the target object in the activity area based on the second monitoring image, and the method comprises the following steps.

[0014] performing image tracking on each target object in the second monitoring image to obtain a tracking trajectory of each target object;

[0015] analyzing the activity amount of each target object based on the tracking trajectory of each target object to obtain the activity amount of each target object.

[0016] Optionally, the tracking trajectory of each target object is obtained by performing image tracking on each target object in the second monitoring image, and the method comprises the following steps.

[0017] performing second target detection on each frame of image in the second monitoring image to obtain a target detection image corresponding to each target object;

[0018] allocating a tracking identifier to the target detection image corresponding to each target object;

[0019] extracting feature points from the target detection image corresponding to each target object to obtain a feature point set corresponding to each target object;

[0020] calculating the spatial position of each target object according to the feature point set corresponding to each target object;

[0021] Based on the spatial position of each target object, a tracking trajectory of each target object is obtained.

[0022] Optionally, the second monitoring image comprises an infrared monitoring image, and after the tracking trajectory of each target object is obtained based on the spatial position of each target object, the method further comprises:

[0023] performing second image segmentation on the target detection image corresponding to each target object to obtain a segmentation image of each target object;

[0024] mapping the segmentation image of each target object into a corresponding infrared frame image in the infrared monitoring image to obtain an infrared image of each target object;

[0025] According to the infrared image of each target object, the tracking trajectory of each target object is marked with a heat map to obtain a heat feature sequence corresponding to the tracking trajectory of each target object.

[0026] Optionally, after the tracking trajectory of each target object is obtained based on the spatial position of each target object, the method further comprises:

[0027] Based on the target detection image corresponding to each target object, action detection is performed on each target object to obtain action information of each target object;

[0028] According to the action information, a corresponding action coefficient is matched;

[0029] According to the tracking trajectory and the action coefficient corresponding to the action information, an action feature sequence corresponding to the tracking trajectory of each target object is obtained.

[0030] Optionally, based on the tracking trajectory of each target object, activity amount analysis is performed on each target object to obtain an activity amount of each target object, comprising:

[0031] Based on the heat feature sequence and / or the action feature sequence of each target object, the tracking trajectory of each target object is combined to perform activity amount analysis on each target object to obtain an activity amount of each target object.

[0032] In a second aspect, an embodiment of the present application provides an automatic feeding device for biological fermentation feed, which comprises:

[0033] an acquisition module configured to acquire a first monitoring image of a feeding area and a second monitoring image of an activity area;

[0034] The identification module is configured to identify the residual amount of the bio-fermented feed in the feeding area based on the first monitoring image.

[0035] The analysis module is configured to analyze the activity amount of the target object in the activity area based on the second monitoring image.

[0036] The processing module is configured to determine the feeding amount of the bio-fermented feed based on the residual amount of the bio-fermented feed and the activity amount of the target object, and feed the bio-fermented feed to the feeding area according to the feeding amount.

[0037] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the steps in the bio-fermented feed automatic feeding method provided by the embodiments of the present application when running the computer program.

[0038] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps in the bio-fermented feed automatic feeding method provided by the embodiments of the present application.

[0039] In the embodiments of the present application, the first monitoring image of the feeding area and the second monitoring image of the activity area are obtained, the residual amount of the bio-fermented feed in the feeding area is identified based on the first monitoring image, the activity amount of the target object in the activity area is analyzed based on the second monitoring image, the feeding amount of the bio-fermented feed is determined based on the residual amount of the bio-fermented feed and the activity amount of the target object, and the bio-fermented feed is fed to the feeding area according to the feeding amount. By identifying the residual amount of the bio-fermented feed in the feeding area through the first monitoring image and analyzing the activity amount of the target object in the activity area through the second monitoring image, the feeding amount of the bio-fermented feed is determined according to the activity amount of the target object and the residual amount of the bio-fermented feed in the feeding area, which can more accurately feed the bio-fermented feed and improve the accuracy of the feeding amount of the target object. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0041] Figure 1is a flow chart of a biological fermentation feed automatic feeding method provided by an embodiment of the present application;

[0042] Figure 2 is a structural schematic diagram of a biological fermentation feed automatic feeding device provided by an embodiment of the present application;

[0043] Figure 3 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0045] Please refer to Figure 1 , Figure 1 is a flow chart of a biological fermentation feed automatic feeding method provided by an embodiment of the present application, as shown in Figure 1 the method comprises the following steps:

[0046] 101, obtaining a first monitoring image of a feeding area and a second monitoring image of an activity area.

[0047] In the embodiments of the present application, the feeding area refers to a biological fermentation feed delivery area, and is also a feeding area of the target object.

[0048] The first monitoring image of the feeding area can be obtained by monitoring the feeding area through the first monitoring camera arranged in the feeding area after the target object finishes eating or after there is no target object in the feeding area. Of course, a video segment without any target object can also be intercepted from the monitoring video of the first monitoring camera as the first monitoring image.

[0049] The target object can be a domesticable animal, such as a chicken, a duck, a pig, a cow, a sheep or the like.

[0050] The activity area is an activity area of the target object, and the activity area comprises one or more second monitoring cameras. The second monitoring image can be obtained by monitoring the activity area through the one or more second monitoring cameras. Specifically, the second monitoring image can be obtained by monitoring the activity area through the second monitoring camera when any target object appears in the activity area. Of course, a video segment with the target object can also be intercepted from the second monitoring camera as the second monitoring image.

[0051] The first monitoring image and the second monitoring image each include multiple frames of continuous frame images.

[0052] 102. Based on the first monitoring image, the residual amount of the biological fermented feed in the feeding area is identified to obtain a biological fermented feed residual amount.

[0053] In the embodiment of the present application, the biological fermented feed residual amount represents the biological fermented feed remaining in the feeding area. The biological fermented feed residual amount can be determined by target detection on the first monitoring image to determine whether the first monitoring image contains the biological fermented feed. If the first monitoring image does not contain the biological fermented feed, it indicates that the biological fermented feed residual amount is 0. If the first monitoring image contains the biological fermented feed, the biological fermented feed residual amount can be determined according to the area of the biological fermented feed. For example, the larger the area of the biological fermented feed, the more the biological fermented feed residual amount, and the smaller the area of the biological fermented feed, the less the biological fermented feed residual amount.

[0054] 103. Based on the second monitoring image, the activity amount of the target object in the activity area is analyzed to obtain the activity amount of the target object.

[0055] In the embodiment of the present application, the target object can be a domesticable animal such as a chicken, a duck, a pig, a cow, a sheep, or the like. The activity amount can be determined according to the activity time and space of the target object, and the activity time and space can be determined by the tracking trajectory of the target object.

[0056] In a possible embodiment, the second monitoring image can be tracked by a target tracking algorithm to obtain the tracking trajectory of each target object in the second monitoring image.

[0057] The longer the tracking trajectory of the target object, the longer the activity time of the target object and the greater the activity amount.

[0058] 104. Based on the biological fermented feed residual amount and the activity amount of the target object, the feeding amount of the biological fermented feed is determined, and the biological fermented feed is fed to the feeding area according to the feeding amount.

[0059] In the embodiment of the present application, for the target object, the relationship between the activity amount and the biological fermented feed intake amount is determined by a relationship table. In the relationship table, different activity amounts correspond to different biological fermented feed intake amounts. The greater the activity amount of a target object, the greater the feeding demand of the target object, and the greater the corresponding biological fermented feed intake amount. Based on the relationship table, the feeding amount of each target object is calculated according to the activity amount of each target object, and the total feeding amount of all target objects is calculated according to the feeding amount of each target object. The feeding amount of the biological fermented feed is obtained by subtracting the biological fermented feed residual amount from the total feeding amount.

[0060] The above relationship table can be set according to expert experience, and after the feeding amount of the biological fermentation feed is obtained, the biological fermentation feed can be fed to the feeding area through an automatic feeding machine or manually.

[0061] In the embodiment of the present application, the first monitoring image of the feeding area and the second monitoring image of the activity area are obtained; the residual amount of the biological fermentation feed in the feeding area is identified based on the first monitoring image, and the residual amount of the biological fermentation feed is obtained; the activity amount of the target object in the activity area is analyzed based on the second monitoring image, and the activity amount of the target object is obtained; and the feeding amount of the biological fermentation feed is determined based on the residual amount of the biological fermentation feed and the activity amount of the target object, and the biological fermentation feed is fed to the feeding area according to the feeding amount. By identifying the residual amount of the biological fermentation feed in the feeding area through the first monitoring image, and analyzing the activity amount of the target object in the activity area through the second monitoring image, and determining the feeding amount of the biological fermentation feed according to the activity amount of the target object and the residual amount of the biological fermentation feed in the feeding area, the feeding of the biological fermentation feed can be more accurate, and the feeding accuracy of the target object can be improved.

[0062] Optionally, in the step of identifying the residual amount of the biological fermentation feed in the feeding area based on the first monitoring image, first target detection can be performed on each frame of image in the first monitoring image to obtain a target image of the feeding area; first image segmentation can be performed on the target image of the feeding area to obtain a background image of the feeding area and a biological fermentation feed image of the feeding area; and the residual amount of the biological fermentation feed in the feeding area can be identified according to the background image of the feeding area and the biological fermentation feed image of the feeding area.

[0063] In the embodiment of the present application, the first monitoring image includes a plurality of continuous frames of images, and first target detection is performed on each frame of image in the first monitoring image, and the detection target of the first target detection is a target image of the feeding area. The feeding area is an area where the biological fermentation feed is put in, and the feeding area can be a feed trough. The target image of the feeding area includes a feed trough as a background and a biological fermentation feed as a foreground.

[0064] When there is no biological fermentation feed in the feeding area, there is no foreground image, and when there is biological fermentation feed in the feeding area, there is a foreground image.

[0065] The first target detection can be based on a target detection algorithm, and the first target detection outputs a detection box (x1, y1, w1, h1, u1) of the feeding area, wherein (x1, y1) is the center coordinate of the detection box of the feeding area, w1 is the width of the detection box of the feeding area, h1 is the height of the detection box of the feeding area, and u1 is the confidence of the detection box of the feeding area. The image region in the detection box of the feeding area is the target image of the feeding area.

[0066] After the target image of the feeding area is obtained, the target image of the feeding area is subjected to first image segmentation to obtain a background image of the feeding area and a bio-fermented feed image of the feeding area. The first image segmentation can be performed based on an image segmentation algorithm, and the first image segmentation outputs a segmented background and a segmented foreground. The foreground is the bio-fermented feed image. The bio-fermented feed remaining amount can be determined by the bio-fermented feed image area. The larger the bio-fermented feed image area is, the more the bio-fermented feed remaining amount is.

[0067] In a possible embodiment, an area ratio between the characteristic fermented feed image area and the background image area of the feeding area can be calculated, and the bio-fermented feed remaining amount can be determined by the area ratio. The larger the area ratio is, the more the bio-fermented feed remaining amount is.

[0068] Optionally, in the step of determining the activity amount of each target object in the activity area based on the second monitoring image, image tracking can be performed on each target object in the second monitoring image to obtain a tracking trajectory of each target object; and the activity amount of each target object can be analyzed based on the tracking trajectory of each target object to obtain the activity amount of each target object.

[0069] In the embodiment of the present application, the second monitoring image includes a plurality of continuous images, and image tracking can be performed on each target object in the second monitoring image by using an image tracking algorithm to obtain a tracking trajectory of each target object.

[0070] The longer the tracking trajectory of the target object is, the longer the activity time of the target object is, and the larger the activity amount of the target object is. The activity amount of each target object can be quantified according to the length of the tracking trajectory of each target object.

[0071] Optionally, in the step of performing image tracking on each target object in the second monitoring image to obtain a tracking trajectory of each target object, second target detection can be performed on each image in the second monitoring image to obtain a target detection image corresponding to each target object; a tracking identifier is assigned to the target detection image corresponding to each target object; feature point extraction is performed on the target detection image corresponding to each target object to obtain a feature point set corresponding to each target object; the spatial position of each target object is calculated according to the feature point set corresponding to each target object; and the tracking trajectory of each target object is obtained based on the spatial position of each target object.

[0072] In the embodiments of the present application, the second target detection is performed based on a target detection algorithm, and a detection frame of each target object in the same frame image is output through the second target detection, wherein the detection frame of the target object is represented by (x2, y2, w2, h2, u2), wherein (x2, y2) is the center coordinate of the detection frame of the target object, w2 is the width of the detection frame of the target object, h2 is the height of the detection frame of the target object, and u2 is the confidence of the detection frame of the target object region, and the image region in the detection frame of the target object is the target detection image corresponding to the target object.

[0073] In a possible embodiment, when each target object appears in the second monitoring image for the first time, a tracking identifier can be assigned to the detection frame of the target object, and the tracking identifier can be a tracking ID. The detection frames of the same target object in different frame images are associated through the tracking ID, and then the target detection images of the same target object in different frame images can be associated. The tracking trajectory of the target object can be obtained by connecting the center coordinates of the detection frames of the same target object as connection points in the time sequence of the frame images. The distance from the connection points to the second monitoring camera is calculated through the camera coordinate system of the second monitoring camera, and thus the positions of the connection points in the real world are obtained. The positions of the connection points in the real world represent the positions of the target objects in the real world, and thus the tracking trajectory of the target object in the real world can be obtained by mapping the tracking trajectory in the second monitoring image to the real world.

[0074] After obtaining the target detection image corresponding to each target object, feature point extraction can be performed on the target detection image corresponding to each target object to obtain a feature point set corresponding to each target object. The spatial position of each target object is calculated according to the feature point set corresponding to each target object, and the spatial position is the geometric center of all feature points in the feature point set. The distance from the geometric center to the second monitoring camera is calculated, and thus the spatial position of the geometric center in the real world is obtained. For a target object, the spatial position of the target object in the real world is connected in the time sequence of the frame images, and thus the tracking trajectory of the target object in the real world is obtained.

[0075] Optionally, the second monitoring image includes an infrared monitoring image. After the step of obtaining the tracking trajectory of each target object based on the spatial position of each target object, a second image segmentation can be performed on the target detection image corresponding to each target object to obtain a segmentation image of each target object. The segmentation image of each target object is mapped into the corresponding infrared frame image in the infrared monitoring image to obtain an infrared image of each target object. The tracking trajectory of each target object is marked with a heat map according to the infrared image of each target object, and thus a heat feature sequence corresponding to the tracking trajectory of each target object is obtained.

[0076] In the embodiment of the present application, the second monitoring image includes an RGB monitoring image and an infrared monitoring image, the second monitoring camera has an infrared image acquisition function, and the second monitoring camera can simultaneously acquire the RGB monitoring image and the infrared monitoring image. The RGB monitoring image is used for second target detection and second image segmentation. The RGB monitoring image and the infrared monitoring image are in one-to-one correspondence.

[0077] The second target detection is performed on each frame image in the RGB monitoring image to obtain a target detection image corresponding to each target object. The second image segmentation is performed on the target detection image corresponding to each target object to obtain a segmentation image of each target object. The segmentation image of the target object includes the contour and position of the target object. The contour and position of the target object are mapped into the corresponding frame image of the infrared monitoring image, and the infrared image of the target object can be obtained. The thermal value corresponding to the target object is obtained according to the infrared image in the contour of the target object.

[0078] Since the infrared images of different target objects are different, the infrared images of the same target object also change with time according to the heat emission condition of the target object. Through the infrared image of the target object, the heat condition of the target object at each position in the tracking process can be obtained. Specifically, the connection point in the tracking trajectory is the position of the target object at a time, and the connection point corresponds to a frame image. The infrared image of the target object in the frame image can be quantified as a thermal value, so that the thermal value is used as a thermal force mark to mark the corresponding connection point in the tracking trajectory, and a thermal feature sequence corresponding to the tracking trajectory of the target object is obtained.

[0079] Through the thermal feature sequence corresponding to the tracking trajectory of the target object, the heat emission condition of each target object at different times and spaces can be obtained.

[0080] Optionally, after obtaining the tracking trajectory of each target object based on the spatial position of each target object, the action detection of each target object can be performed based on the target detection image corresponding to each target object to obtain the action information of each target object. According to the action information, the corresponding action coefficient is matched. According to the action coefficient corresponding to the tracking trajectory and the action information, an action feature sequence corresponding to the tracking trajectory of each target object is obtained.

[0081] In the embodiment of the present application, the second target detection is performed on each frame image in the RGB monitoring image to obtain a target detection image corresponding to each target object. The action detection of each target object can be performed to obtain the action information of each target object. Since different actions represent different activity amounts, the action detection of the target object can better obtain the activity amount of the target object.

[0082] Different motion information corresponds to different motion systems. For example, the motion coefficient for walking is R, the motion coefficient for running can be AR, and the motion coefficient for jumping can be BR, etc., where A and B are both numbers greater than 1.

[0083] By using the action information of the target object, we can obtain the action of the target object at each position during the tracking process. Specifically, the connection point in the tracking trajectory is the position of the target object at a certain time. The connection point corresponds to a frame of image. The action information of the target object in the frame of image can be matched with the corresponding action coefficient. Thus, the action coefficient of the target object is associated with the corresponding connection point in the tracking trajectory to obtain the action feature sequence corresponding to the tracking trajectory of the target object.

[0084] By analyzing the action feature sequence corresponding to the tracking trajectory of the target object, we can obtain the action status of each target object in different times and spaces, and thus obtain the specific activity status of each target object in different times and spaces.

[0085] Optionally, in the step of performing activity analysis on each target object based on the tracking trajectory of each target object to obtain the activity amount of each target object, the activity analysis can be performed on each target object based on the thermal characteristic sequence and / or motion characteristic sequence of each target object, combined with the tracking trajectory of each target object, to obtain the activity amount of each target object.

[0086] In this embodiment of the invention, the longer the tracking trajectory of the target object, the longer the activity time and the greater the activity of the target object. The thermal feature sequence can represent the heat dissipation of the target object in different times and spaces. The action feature sequence can represent the actions of each target object in different times and spaces, and thus represent the specific activities of each target object in different times and spaces.

[0087] In one possible embodiment, the tracking trajectory is formed by connecting connection points. Essentially, the connection points are connected in the time sequence corresponding to the frame images. Therefore, the tracking trajectory can be represented by the connection point sequence an, which characterizes the spatiotemporal features of the target object.

[0088] The connection point sequence an and the thermal feature sequence bn have the same dimension, where n represents the number of frames in the second monitoring image. The connection point sequence an and the thermal feature sequence bn can be fused by channels to obtain the first fused feature cn. The first fused feature cn is then input into the preset first prediction model to output the predicted delivery amount.

[0089] The connection point sequence an has the same dimension as the action feature sequence dn, n represents that there are n frame images in the second monitoring image, the connection point sequence an and the thermal feature sequence dn can be channel fused to obtain a second fusion feature en, and the second fusion feature en is input into a preset second prediction model to output the predicted delivery amount.

[0090] The connection point sequence an, the thermal feature sequence bn and the action feature sequence dn have the same dimension, n represents that there are n frame images in the second monitoring image, the connection point sequence an, the thermal feature sequence bn and the action feature sequence dn can be channel fused to obtain a third fusion feature fn, and the third fusion feature fn is input into a preset third prediction model to output the predicted delivery amount.

[0091] The prediction model is trained, the sample connection point sequence am, the sample thermal feature sequence bm and the action feature sequence dm can be extracted from the historical monitoring image, each group of the sample connection point sequence am, the sample thermal feature sequence bm and the action feature sequence dm corresponds to a delivery amount label, the connection point sequence am and the sample thermal feature sequence bm are channel fused to obtain a first sample fusion feature cm, the sample connection point sequence am and the action feature sequence dm are channel fused to obtain a second sample fusion feature em, and the sample connection point sequence am, the sample thermal feature sequence bm and the action feature sequence dm are channel fused to obtain a third sample fusion feature fm. The prediction model is supervised trained through the first sample fusion feature cm and the corresponding label, the first prediction model can be trained, the prediction model is supervised trained through the second sample fusion feature em and the corresponding label, the second prediction model can be trained, and the prediction model is supervised trained through the third sample fusion feature fm and the corresponding label, the third prediction model can be trained.

[0092] It should be noted that the first prediction model, the second prediction model and the third prediction model are different in input data, and the parameters of the prediction models are different, the delivery amount is predicted through different prediction models, and the flexibility of prediction can be increased.

[0093] It should be noted that the biological fermentation feed automatic feeding method provided by the embodiment of the present application can be applied to intelligent feeding machines, smart phones, computers, servers and the like.

[0094] Optionally, the present embodiment provides a biological fermentation feed automatic feeding device, please refer to Figure 2 , Figure 2 It is a structure schematic diagram of a biological fermentation feed automatic feeding device provided by the embodiment of the present application, as Figure 2 shown, the device comprises:

[0095] The acquisition module 201 is configured to acquire a first monitoring image of a feeding area and a second monitoring image of an activity area.

[0096] The identification module 202 is configured to perform residual amount identification on biological fermented feed in the feeding area based on the first monitoring image, to obtain a biological fermented feed residual amount.

[0097] The analysis module 203 is configured to perform activity amount analysis on a target object in the activity area based on the second monitoring image, to obtain an activity amount of the target object.

[0098] The processing module 204 is configured to determine a feeding amount of the biological fermented feed based on the biological fermented feed residual amount and the activity amount of the target object, and perform feeding of the biological fermented feed to the feeding area according to the feeding amount.

[0099] Optionally, the identification module 202 is further configured to perform first target detection on each frame of image in the first monitoring image, to obtain a target image of the feeding area; perform first image segmentation on the target image of the feeding area, to obtain a background image of the feeding area and a biological fermented feed image of the feeding area; and identify the biological fermented feed residual amount in the feeding area according to the background image of the feeding area and the biological fermented feed image of the feeding area.

[0100] Optionally, the analysis module 203 is further configured to perform image tracking on each target object in the second monitoring image, to obtain a tracking trajectory of each target object; and perform activity amount analysis on each target object based on the tracking trajectory of each target object, to obtain an activity amount of each target object.

[0101] Optionally, the analysis module 203 is further configured to perform second target detection on each frame of image in the second monitoring image, to obtain a target detection image corresponding to each target object; assign a tracking identifier to the target detection image corresponding to each target object; perform feature point extraction on the target detection image corresponding to each target object, to obtain a feature point set corresponding to each target object; calculate a spatial position of each target object according to the feature point set corresponding to each target object; and obtain a tracking trajectory of each target object based on the spatial position of each target object.

[0102] Optionally, the analysis module 203 is further configured to perform second image segmentation on each target detection image corresponding to each target object to obtain a segmented image of each target object; map the segmented image of each target object into a corresponding infrared frame image in the infrared monitoring image to obtain an infrared image of each target object; and perform heat map marking on the tracking trajectory of each target object according to the infrared image of each target object to obtain a heat feature sequence corresponding to the tracking trajectory of each target object.

[0103] Optionally, the analysis module 203 is further configured to perform action detection on each target object based on the target detection image corresponding to each target object to obtain action information of each target object; match a corresponding action coefficient according to the action information; and obtain an action feature sequence corresponding to the tracking trajectory of each target object according to the tracking trajectory and the action coefficient corresponding to the action information.

[0104] Optionally, the analysis module 203 is further configured to perform activity amount analysis on each target object based on the heat feature sequence and / or the action feature sequence of each target object in combination with the tracking trajectory of each target object to obtain an activity amount of each target object.

[0105] It should be noted that the behavior detection device provided by the embodiment of the present application can be applied to intelligent feeding machines, smart phones, computers, servers and other devices that can automatically feed biological fermentation feed.

[0106] The behavior detection device provided by the embodiment of the present application can realize each process of the biological fermentation feed automatic feeding method in the above method embodiment, and can achieve the same beneficial effects. To avoid repetition, it will not be repeated here.

[0107] Referring to Figure 3 , Figure 3 is a structural schematic diagram of an electronic device provided by the embodiment of the present application, as Figure 3 shown, comprising: a memory 302, a processor 301 and a computer program of the biological fermentation feed automatic feeding method stored in the memory 302 and executable on the processor 3401, wherein:

[0108] The processor 301 is configured to call the computer program stored in the memory 302 to perform the following steps:

[0109] obtain a first monitoring image of a feeding area and a second monitoring image of an activity area;

[0110] perform residual amount identification on the biological fermentation feed in the feeding area based on the first monitoring image to obtain a biological fermentation feed residual amount;

[0111] perform activity amount analysis on the target object in the active area based on the second monitoring image to obtain the activity amount of the target object;

[0112] determine the feeding amount of the bio-fermented feed based on the bio-fermented feed remaining amount and the activity amount of the target object, and perform feeding of the bio-fermented feed to the feeding area according to the feeding amount.

[0113] Optionally, the processor 301 performs the bio-fermented feed remaining amount identification of the feeding area based on the first monitoring image to obtain the bio-fermented feed remaining amount, including:

[0114] perform first target detection on each frame of image in the first monitoring image to obtain a target image of the feeding area;

[0115] perform first image segmentation on the target image of the feeding area to obtain a background image of the feeding area and a bio-fermented feed image of the feeding area;

[0116] identify the bio-fermented feed remaining amount in the feeding area according to the background image of the feeding area and the bio-fermented feed image of the feeding area.

[0117] Optionally, the processor 301 performs the activity amount analysis on the target object in the active area based on the second monitoring image to obtain the activity amount of the target object, including:

[0118] perform image tracking on each of the target objects in the second monitoring image to obtain a tracking trajectory of each of the target objects;

[0119] perform activity amount analysis on each of the target objects based on the tracking trajectory of each of the target objects to obtain the activity amount of each of the target objects.

[0120] Optionally, the processor 301 performs the image tracking on each of the target objects in the second monitoring image to obtain the tracking trajectory of each of the target objects, including:

[0121] perform second target detection on each frame of image in the second monitoring image to obtain a target detection image corresponding to each of the target objects;

[0122] assign a tracking identifier to the target detection image corresponding to each of the target objects;

[0123] perform feature point extraction on the target detection image corresponding to each of the target objects to obtain a feature point set corresponding to each of the target objects;

[0124] According to the feature point set corresponding to each target object, a spatial position of each target object is calculated;

[0125] Based on the spatial position of each target object, a tracking trajectory of each target object is obtained.

[0126] Optionally, the second monitoring image includes an infrared monitoring image, and after the tracking trajectory of each target object is obtained based on the spatial position of each target object, the method executed by the processor 301 further includes:

[0127] A second image segmentation is performed on the target detection image corresponding to each target object to obtain a segmentation image of each target object;

[0128] The segmentation image of each target object is mapped into the corresponding infrared frame image in the infrared monitoring image to obtain an infrared image of each target object;

[0129] According to the infrared image of each target object, a heat map is marked on the tracking trajectory of each target object to obtain a heat feature sequence corresponding to the tracking trajectory of each target object.

[0130] Optionally, after the tracking trajectory of each target object is obtained based on the spatial position of each target object, the method executed by the processor 301 further includes:

[0131] Based on the target detection image corresponding to each target object, an action detection is performed on each target object to obtain action information of each target object;

[0132] According to the action information, a corresponding action coefficient is matched;

[0133] According to the tracking trajectory and the action coefficient corresponding to the action information, an action feature sequence corresponding to the tracking trajectory of each target object is obtained.

[0134] Optionally, the activity amount analysis on each target object based on the tracking trajectory of each target object executed by the processor 301 includes:

[0135] Based on the heat feature sequence and / or the action feature sequence of each target object, the activity amount analysis on each target object is performed in combination with the tracking trajectory of each target object to obtain the activity amount of each target object.

[0136] It should be noted that the electronic device provided by the embodiment of the present application can be applied to a smart phone, a computer, a server and the like which can automatically feed the bio-fermented feed.

[0137] The electronic device provided by the embodiment of the present application can realize each process of the bio-fermented feed automatic feeding method in the above-mentioned method embodiment, and can achieve the same beneficial effects. To avoid repetition, details are not described here.

[0138] The embodiment of the present application also provides a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to realize each process of the bio-fermented feed automatic feeding method or the application end bio-fermented feed automatic feeding method provided by the embodiment of the present application, and can achieve the same technical effects. To avoid repetition, details are not described here.

[0139] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by a computer program to instruct related hardware. The program can be stored in a computer readable storage medium. When the program is executed, the program can include the processes of the above-mentioned embodiment of each method. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM) and the like.

[0140] The above only describes the preferred embodiments of the present application, and of course cannot limit the scope of the rights of the present application, so the equivalent changes made according to the claims of the present application still belong to the scope covered by the present application.

Claims

1. A method for automatically feeding a bio-fermented feed, characterized by, The method comprises the following steps: acquiring a first monitoring image of a feeding area and a second monitoring image of an activity area; based on the first monitoring image, identifying the remaining amount of bio-fermented feed in the feeding area to obtain the bio-fermented feed remaining amount; based on the second monitoring image, analyzing the activity amount of a target object in the activity area, including: tracking each target object in the second monitoring image to obtain the tracking trajectory of each target object; performing second target detection on each frame of image in the second monitoring image to obtain the target detection image corresponding to each target object; assigning a tracking identifier to the target detection image corresponding to each target object; extracting feature points from the target detection image corresponding to each target object to obtain the feature point set corresponding to each target object; calculating the spatial position of each target object according to the feature point set corresponding to each target object; obtaining the tracking trajectory of each target object based on the spatial position of each target object; performing second image segmentation on the target detection image corresponding to each target object to obtain the segmentation image of each target object; mapping the segmentation image of each target object to the corresponding infrared frame image in the infrared monitoring image to obtain the infrared image of each target object; based on the infrared image of each target object, marking the tracking trajectory of each target object with a heat map to obtain the heat feature sequence corresponding to the tracking trajectory of each target object; based on the target detection image corresponding to each target object, detecting the action of each target object to obtain the action information of each target object; matching the corresponding action coefficient according to the action information; obtaining the action feature sequence corresponding to the tracking trajectory of each target object according to the tracking trajectory and the action coefficient corresponding to the action information; based on the heat feature sequence and the action feature sequence of each target object, and in combination with the tracking trajectory of each target object, analyzing the activity amount of each target object to obtain the activity amount of each target object; based on the bio-fermented feed remaining amount and the activity amount of the target object, determining the feeding amount of the bio-fermented feed, and feeding the bio-fermented feed to the feeding area according to the feeding amount.

2. The method of claim 1, wherein, The method comprises the following steps: performing first target detection on each frame of image in the first monitoring image to obtain the target image of the feeding area; performing first image segmentation on the target image of the feeding area to obtain the background image of the feeding area and the bio-fermented feed image of the feeding area; based on the background image of the feeding area and the bio-fermented feed image of the feeding area, identifying the bio-fermented feed remaining amount in the feeding area.

3. An automatic feeding device for bio-fermented feed, characterized in that, The device comprises: The acquisition module is configured to acquire a first monitoring image of a feeding area and a second monitoring image of an activity area; The identification module is configured to perform residual amount identification on the bio-fermented feed in the feeding area based on the first monitoring image to obtain a bio-fermented feed residual amount; The analysis module is configured to perform activity amount analysis on a target object in the activity area based on the second monitoring image to obtain an activity amount of the target object, including: performing image tracking on each target object in the second monitoring image to obtain a tracking trajectory of each target object; performing second target detection on each frame of image in the second monitoring image to obtain a target detection image corresponding to each target object; allocating a tracking identifier to the target detection image corresponding to each target object; performing feature point extraction on the target detection image corresponding to each target object to obtain a feature point set corresponding to each target object; calculating a spatial position of each target object according to the feature point set corresponding to each target object; obtaining a tracking trajectory of each target object based on the spatial position of each target object; performing second image segmentation on the target detection image corresponding to each target object to obtain a segmentation image of each target object; mapping the segmentation image of each target object into a corresponding infrared frame image in an infrared monitoring image to obtain an infrared image of each target object; performing heat map marking on the tracking trajectory of each target object according to the infrared image of each target object to obtain a heat feature sequence corresponding to the tracking trajectory of each target object; performing action detection on each target object based on the target detection image corresponding to each target object to obtain action information of each target object; matching a corresponding action coefficient according to the action information; obtaining an action feature sequence corresponding to the tracking trajectory of each target object according to the tracking trajectory and the action coefficient corresponding to the action information of each target object; performing activity amount analysis on each target object based on the heat feature sequence and the action feature sequence of each target object in combination with the tracking trajectory of each target object to obtain an activity amount of each target object; The processing module is configured to determine a feeding amount of the bio-fermented feed based on the bio-fermented feed residual amount and the activity amount of the target object, and to feed the bio-fermented feed to the feeding area according to the feeding amount.

Images