Pet feeder control method and system, terminal equipment and storage medium

Abstract

The invention belongs to the technical field of pet supplies, and discloses a pet feeder control method and system, terminal equipment and a storage medium, and the method comprises the steps: obtaining a target image in a target region in front of a pet feeder in real time; when it is not detected that the target animal exists in the target area through the target image, gesture action recognition is conducted on the target image through a gesture recognition model; when the gesture action is recognized as the target gesture, the pet feeder is controlled to execute a corresponding target instruction, so that a motor of the pet feeder is controlled to be turned on or turned off. After the target image in the target area in front of the pet feeder is acquired in real time, the target animal in the target image is detected, the gesture action is recognized when the target image is not detected, and when the target gesture is recognized to exist, the motor of the pet feeder is controlled to be turned on or turned off so as to correspondingly control whether to release food or not. Automatic feeding controlled by gestures is achieved, and convenience and flexibility are greatly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pet products, in particular to a pet feeder control method and system, a terminal device and a storage medium. BACKGROUND

[0002] Traditional pet feeders mainly use physical buttons, timing and fixed quantity or remote APP control methods, which have the following disadvantages: Physical buttons may be dirty, and user operation is not hygienic; timing feeding cannot be flexibly adjusted according to the actual needs of pets; and remote control requires users to always carry mobile devices, which is inconvenient to operate.

[0003] Therefore, the prior art still needs to be improved and improved. SUMMARY

[0004] The present application provides a pet feeder control method and system, a terminal device and a storage medium, which aims to solve the problem of low convenience and poor flexibility in controlling pet feeders to feed in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a pet feeder control method, comprising: real-time acquisition of a target image in a target area in front of a pet feeder; when no target animal is detected in the target area through the target image, performing gesture action recognition on the target image using a gesture recognition model; when the gesture action is identified as a target gesture, controlling the pet feeder to execute a corresponding target instruction to control the pet feeder to start or stop the motor.

[0006] In some embodiments, the gesture action in the target area is recognized using a gesture recognition model, and then further comprising: when the gesture action is not recognized, or when the gesture action is not identified as the target gesture, continue to acquire the target image of the target area in front of the pet feeder.

[0007] In some embodiments, the target image in the target area in front of the pet feeder is acquired in real time, and then further comprising: differential operation is performed on the current target image and a plurality of historical target images, and whether the current target image changes is determined according to the operation result; when the current target image changes, it is determined that the target animal exists in the current target image, and the gesture action is stopped from being recognized; The plurality of historical target images are a continuous preset number of images before the current target image.

[0008] In some embodiments, the controlling the pet feeder to execute the out-of-food instruction further comprises: controlling the pet feeder to stop executing the out-of-food instruction when the target image detects that there is a target animal in the target area or the food at the outlet of the pet feeder exceeds a preset weight.

[0009] In some embodiments, when the gesture action is identified as a target gesture, the controlling the pet feeder to execute a corresponding target instruction comprises: controlling the pet feeder to execute an out-of-food instruction to start the motor to release a preset amount of food when the gesture action is identified as a first target gesture; controlling the pet feeder to stop executing the out-of-food instruction to stop releasing the food when the gesture action is identified as a second target gesture; wherein the first target gesture is a fist gesture and the second target gesture is a flat gesture.

[0010] In some embodiments, the preset amount is calculated by an out-of-food amount control formula; wherein the out-of-food amount control formula is: ; wherein Q is the out-of-food amount, N is the number of motor rotations, d is the diameter of the screw propeller in the pet feeder, L is the lead of the screw, and p is the bulk density of the food.

[0011] In some embodiments, the identifying the gesture action in the target area using the gesture recognition model further comprises: after collecting a plurality of gesture sample data, performing data enhancement processing on the gesture sample data to obtain processed data; training the processed data using a classification loss function and a bounding box regression loss function, and obtaining the gesture recognition model when a target number of iterations is reached.

[0012] In a second aspect, the embodiments of the present application provide a pet feeder control system, comprising: a detection module configured to acquire a target image in a target area in front of a pet feeder in real time; an identification module configured to identify a gesture action in the target image using a gesture recognition model when the target image does not detect that there is a target animal in the target area; a control module configured to control the pet feeder to execute a corresponding target instruction when the gesture action is identified as a target gesture, so as to control the pet feeder to start or stop the motor.

[0013] In a third aspect, an embodiment of the present application provides a pet feeder, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and when the computer program is executed by the processor, the steps of the pet feeder control method described above are implemented.

[0014] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of the pet feeder control method described above are implemented.

[0015] Compared with the prior art, the pet feeder control method, system, terminal device and storage medium provided by the present application can realize automatic feeding controlled by gestures, greatly improve the convenience and flexibility, by acquiring a target image in a target area in front of the pet feeder in real time, detecting a target animal in the target image, recognizing a gesture action when the target animal is not detected, and controlling the motor of the pet feeder to start or stop when the target gesture is recognized. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0017] Figure 1 A flow chart of the pet feeder control method provided by the present application; Figure 2 A flow chart of the pet feeder control method provided by the present application; Figure 3 An illustration of recognizing the presence of a target animal in the pet feeder control method provided by the present application; Figure 4 An illustration of training a gesture recognition model in the pet feeder control method provided by the present application; Figure 5 An illustration of recognizing a gesture action in the pet feeder control method provided by the present application; Figure 6 An illustration of a fist gesture in the pet feeder control method provided by the present application; Figure 7 An illustration of a flat hand gesture in the pet feeder control method provided by the present application; Figure 8 An illustration of executing a target instruction in the pet feeder control method provided by the present application; Figure 9 A structural schematic diagram of a pet feeder control system provided by the present application.

[0018] Reference signs: 10-detection module; 20-identification module; 30-control module. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application.

[0020] The components of the embodiments of the present application generally described and illustrated in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0021] Hereinafter, the terms “include”, “have”, and their conjugates used in the various embodiments of the present application are only intended to denote that specific features, numbers, steps, operations, elements, components, or combinations thereof are present, and should not be understood as excluding the presence or addition of one or more other features, numbers, steps, operations, elements, components, or combinations thereof. In addition, the terms “first”, “second”, “third”, and the like are only used to distinguish descriptions, and cannot be understood as indicating or implying relative importance.

[0022] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the present application belong. The terms (such as those defined in commonly used dictionaries) will be interpreted as having a meaning that is the same as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the present application.

[0023] The application provides a pet feeder control method, system, terminal equipment and storage medium. The pet feeder control method detects a target animal in a target image after acquiring the target image in a target area in front of a pet feeder in real time, identifies a gesture action when the target animal is not detected, executes a target instruction corresponding to the target gesture when the target gesture is identified, and controls a motor of the pet feeder to start or stop to correspondingly control whether to release food, thereby realizing humanized and intelligent feeding of gesture interaction control and pet position detection, and greatly improving convenience and flexibility.

[0024] The pet feeder control method design scheme is described below through some specific embodiments.

[0025] Please refer to Figure 1 The application embodiment provides a pet feeder control method, which includes steps S100-S300. S100, acquiring a target image in a target area in front of a pet feeder in real time.

[0026] Exemplarily, referring to Figure 2 Before controlling the pet feeder, the parameters of the background difference algorithm in the system and the gesture recognition model are initialized, the motor for releasing food is reset to zero, and an image acquisition module (such as a USB camera or a MIPI (Mobile Industry Processor Interface) interface camera) is started and can be controlled to capture image frames at a period of 66 ms (15 fps).

[0027] Then, the control process is entered: the target image (image captured by the camera) in the target area in front of the pet feeder (for example, a range of 10 cm in front of the pet feeder) is acquired in real time by using the image acquisition module, the pet position can be detected by using the background difference method combined with the detection method of deep learning, and the gesture can be detected by using the image recognition algorithm, so as to detect whether there is a pet or / and a gesture in the target image.

[0028] It can be understood that the target area in front of the pet feeder is imaged in the application to quickly identify the gesture.

[0029] In one implementation method, referring to Figure 3 Step S100, acquiring a target image in a target area in front of a pet feeder in real time, and then further including: S101, performing difference operation on the current target image and multiple historical target images, and judging whether the current target image changes according to the operation result; S102, when the current target image changes, it is judged that there is a target animal in the current target image, and the gesture action recognition is stopped.

[0030] The multiple historical target images are a continuous preset number of images before the current target image.

[0031] Exemplarily, when the detection method of background difference combined with deep learning is used to detect the pet position in the target region: First, foreground matching detection is performed, that is, the current frame image in the target image is subjected to difference operation with the continuous preset number of images before the current target image, that is, the current frame image is subjected to difference operation with the background image in the historical collected continuous multiple (for example, 10) images containing the target pet, so as to calculate the difference image , that is, the difference value between each pixel point and the background.

[0032] The formula of the pixel-level difference operation (or called foreground detection) is: (1); In formula (1), : the current frame image; : the background model; : the dynamic threshold (for example, τ = 25). If the difference is large, it means that the pixel may belong to the foreground.

[0033] Then, it is judged whether the target image changes according to the operation result: that is, whether the target animal exists is judged by using the pet existence determination condition.

[0034] The pet existence determination condition is: (2); In formula (2), R is the ROI region (that is, the target region), is the detected pet vertical coordinate.

[0035] When the background difference result is greater than the threshold , it is judged that the target image changes, and if is also within the set range , it is judged that the target animal exists in the target image, so that the system has the pet position detection function, the system enters the standby protection state, and all gesture inputs are shielded.

[0036] Further, in addition to the above-mentioned foreground region matching method for determining whether an animal exists, any one of the following methods can be used to determine whether an animal exists in the target region: target detection, motion behavior feature, persistence judgment, and behavior semantic feature judgment.

[0037] In the target detection, the acquired target image is input into a deep learning model, and whether the target is a detected "pet" category (such as a cat or a dog) is determined according to the probability output by the deep learning model. The motion behavior feature refers to whether the target has a certain motion feature (such as displacement and speed) within a certain time. The persistence judgment refers to whether the target is detected in consecutive multiple frames of target images, thereby avoiding false detection. The behavior semantic feature judgment refers to determining whether the target conforms to the typical behavior pattern of a pet according to the specific behavior (such as walking, jumping, or being still) of the pet output by the deep learning model.

[0038] Further, in another embodiment of the present application, an infrared distance measuring module can be used to detect whether a target animal exists in the target region, and a deep learning model can be used to determine whether a "pet" category (such as a cat or a dog) is detected. The detection range of the infrared distance measuring module can be 5-50 cm, and the accuracy can be ±1 cm.

[0039] S200, when it is determined that no target animal exists in the target region according to the target image, a gesture recognition model is used to recognize the gesture action in the target image.

[0040] For example, when detecting the target image in the target region, if no target animal exists in the target region in the target image (for example, no object is detected, or only a gesture action is detected), a gesture recognition mode is entered, that is, the target image is input into a gesture recognition model, a predicted gesture category probability is output, and it is determined whether the gesture action is a target gesture (i.e. Figure 2 an effective gesture in the gesture category, including, for example, a fist or an open hand, etc.) according to the gesture category probability, so as to control whether the pet feeder automatically feeds according to the determination result, thereby realizing a double detection mechanism (gesture + pet position) and effectively preventing false triggering.

[0041] It can be understood that, in the present application, the gesture recognition model is used to recognize the gesture action, so as to control the pet feeder to feed according to the determination result, thereby realizing accurate control of the pet feeder to feed according to the gesture action, and effectively improving the feeding efficiency and convenience. In addition, the use of a dedicated gesture recognition model can improve the recognition accuracy and adapt to different user habits.

[0042] In one embodiment, referring to Figure 4 the gesture recognition model is used to recognize the gesture action in the target region, which further includes: S198. After collecting multiple gesture sample data, perform data augmentation processing on the gesture sample data to obtain processed data. S199. Using classification loss function and bounding box regression loss function, train the model on the processed data, and stop training when the target number of iterations is reached to obtain the gesture recognition model. Alternatively, training can be stopped when the accuracy on the validation set no longer improves to prevent overfitting.

[0043] As an example, the training process for a gesture recognition model is as follows: First, after users register with the system, they enter the data preparation phase: this involves collecting multiple gesture sample datasets (e.g., no fewer than 50 samples) and then labeling the gesture bounding boxes and category labels. Next, data augmentation is performed on the gesture sample data. For example, random rotation (rotation angle θ follows a uniform distribution, i.e., θ ~ U(-15°, 15°)), random cropping, and brightness adjustment (brightness adjustment parameter Δβ follows a normal distribution, i.e., Δβ ~ N(0, 0.1)) are used to improve the model's generalization ability, resulting in processed data.

[0044] Random rotation refers to transforming the pixel matrix of the input image around its center point. Each time, the rotation angle θ is independently sampled from a uniform distribution U(-15°, 15°), meaning any angle between -15° and 15° has an equal probability of occurrence. Brightness adjustment is a global modification of the overall brightness of the image. The brightness adjustment parameter Δβ~N(0, 0.1) indicates that each time an adjustment is made, the system randomly samples an offset from a normal distribution with a mean of 0 and a standard deviation of 0.1, applying the same magnitude of brightness change to all pixels.

[0045] Then, we move on to the data model building phase: The user-specific gesture recognition model is trained using architectures such as YOLOv5s (a single-stage detection model in the YOLO series), YOLOv8, and Faster R-CNN (a two-stage detector) as the base models.

[0046] Next, the loss function in the model is set: Replace the original classification loss function (such as cross-entropy loss) with Focal Loss, and retain the original bounding box regression loss (such as CIoU Loss). You can also retain the target confidence loss (such as BCE Loss) at the same time.

[0047] The classification loss function used is Focal Loss, and its calculation formula is as follows: (3); In formula (3), t is the predicted probability; αt is the class balance factor; and γ is the adjustable focusing parameter (default γ=2).

[0048] The bounding box regression loss function can adopt CIoU Loss, and the calculation formula is: (4); In formula (4), p: Euclidean distance, corresponding to the center points of the predicted box and the real box; c: diagonal length of the minimum circumscribed rectangle; v: consistency measurement of aspect ratio.

[0049] Further, after calculating the classification loss function and the bounding box regression loss function, a weighted sum function of the two functions is calculated, and the weighted sum function is simulated training, and the model training iteration number N is set to minimize the total loss function (weighted sum function) to achieve the best performance on the test set, wherein the iteration number is obtained by formula: (5); In formula (5), a is the target confidence (default 0.95), and e is the single iteration promotion rate.

[0050] The classification loss function is used to alleviate the imbalance between positive and negative samples, and can reduce the weight of simple negative samples in training.

[0051] It can be understood that in the present application, the collected gesture sample data is processed by data enhancement, and the processed data is trained by using the classification loss function and the bounding box regression loss function, and when the target iteration number is reached, the gesture recognition model is trained to quickly and accurately recognize the target gesture in the image.

[0052] In one embodiment, referring to Figure 5 , the gesture recognition model is used to recognize the gesture action in the target region, comprising: S201, using the gesture recognition model to analyze the target image with gestures to obtain a plurality of gesture category probabilities; S202, when the target gesture category probability is greater than a preset value, the target gesture corresponding to the target gesture category probability is recognized in the target region.

[0053] Exemplarily, the specific process of recognizing the gesture action is as follows: First, after obtaining the target image, the target image with gesture action is screened out, and the target image with gesture action is input into the gesture recognition model for analysis, and a plurality of gesture category probabilities are output, that is, a plurality of gesture category confidences are output. Among them, the target image with gesture action refers to the same gesture appearing in the target image for more than a preset time (for example, 500ms), so as to realize anti-shake confirmation, thereby effectively avoiding false triggering.

[0054] Then, when a target gesture category probability in the plurality of gesture category probabilities is greater than a preset value (e.g., 0.95), it is determined that a target gesture corresponding to the target gesture category probability occurs, and it is determined that a valid gesture, including a fist gesture and a flat gesture, occurs.

[0055] In an implementation method, the gesture action in the target region is identified by using the gesture recognition model, and then further comprising: When the gesture action is not identified, or when the gesture action is not the target gesture, the target image of the target region in front of the pet feeder is continuously acquired.

[0056] Exemplarily, when it is detected by using the gesture recognition model that the target animal and the gesture action do not exist in the target image, i.e., the gesture action is not identified, the target image in the target region in front of the pet feeder is continuously acquired in real time by using the image acquisition module.

[0057] And, when the gesture action is not the target gesture, i.e., the plurality of gesture category probabilities are all less than the preset value, i.e., there is no corresponding target gesture category, it belongs to an invalid gesture action (including all other gestures except the fist gesture and the flat gesture), at this time, other gestures are ignored, and the target image in the target region is also continuously acquired by using the image acquisition module.

[0058] S300, when the gesture action is identified as the target gesture, the pet feeder is controlled to execute a corresponding target instruction to control the motor of the pet feeder to be turned on or off.

[0059] Exemplarily, after it is identified that the gesture action exists in the target image, it is further identified whether the gesture action is the target gesture: When the gesture action is identified as the target gesture such as the fist gesture or the flat gesture, the pet feeder is controlled to execute a corresponding target instruction to control the motor of the pet feeder to be turned on or off, for example, when the target gesture is the fist, the motor of the pet feeder is controlled to be turned on for a certain time, and the motor controls the spiral propeller at the bottom to rotate, so that the food falls from the food outlet to release a certain amount of food; when the target gesture is the flat gesture, the motor of the pet feeder is controlled to be turned off. Wherein, the fist gesture is shown in Figure 6 , wherein, Figure 6 In the figure, a is the original schematic diagram of the fist gesture, and b is the schematic diagram of the fist gesture after differential identification. The schematic diagram of the flat gesture is shown in Figure 7 , wherein, Figure 7 In the figure, a is the original schematic diagram of the flat gesture, and b is the schematic diagram of the flat gesture after differential identification.

[0060] It can be understood that, in the present application, the motor is turned on or off by identifying the target gesture to achieve automatic and rapid grain release through gesture control, thereby improving convenience. Moreover, a user-specific gesture training model is used to improve recognition accuracy. Furthermore, compared with other control methods such as voice or APP, gesture control is more in line with human natural interaction habits.

[0061] In one implementation method, please refer to Figure 8 When the gesture action is identified as a target gesture, the pet feeder is controlled to execute a corresponding target instruction, including: S301, when the gesture action is identified as a first target gesture, the pet feeder is controlled to execute a grain release instruction to turn on the motor to release a preset amount of grain; S302, when the gesture action is identified as a second target gesture, the pet feeder is controlled to stop executing the grain release instruction to stop releasing grain.

[0062] Exemplarily, the specific steps of controlling the execution of the corresponding target instruction are as follows: When identifying whether the gesture action is a target gesture: If the gesture action is identified as a fist (i.e., the first target gesture), i.e., the probability of the fist gesture corresponding to the target gesture category is greater than a preset value, at this time, the pet feeder is controlled to execute the grain release instruction to control the motor (such as a stepper motor) at the bottom of the pet feeder to rotate to drive the screw propeller to rotate, so as to release a preset amount of grain at a time, wherein the release times can be free.

[0063] Wherein, the preset amount is calculated by the grain release amount control formula; wherein, the grain release amount control formula: (6); Wherein, in formula (6), Q is the grain release amount (g), N is the number of motor rotations, d is the diameter of the screw propeller in the pet feeder (mm), L is the screw lead (mm), and p is the grain bulk density (g / mm ).

[0064] Moreover, in another embodiment of the present application, the preset amount can also be controlled by the following grain release control model, and the grain release amount calculation formula is: (7); Wherein, in formula (7), F is the single grain release amount; v is the grain release speed (for example, a typical value is 5g / s); and ∫dt is the duration integral of the grain release signal. The meaning of the grain release amount calculation formula is: grain release amount = grain release time * grain release speed, and there is a safety limit condition, i.e., the total grain release amount is limited.

[0065] Then, if the gesture action is identified as the flattening gesture (i.e., the second target gesture), i.e., the probability of the target gesture corresponding to the flattening gesture is greater than a preset value, at this time, the pet feeder is controlled to execute the out-of-food instruction to control the motor to stop, so as to close the motor.

[0066] In an implementation method, after the pet feeder executes the out-of-food instruction, further comprising: When the target image detects that there is a target animal in the target area, or the food at the outlet of the pet feeder exceeds the preset weight, the pet feeder is controlled to stop executing the out-of-food instruction.

[0067] Exemplarily, in the pet feeder out-of-food operation, there will be emergency stop conditions, including: pet approaching, stop gesture (i.e., flattening gesture) and out-of-food over-limit. The out-of-food over-limit is a safety limit condition of the food amount, that is, the food at the outlet of the pet feeder is weighed in real time by using the weighing sensor at an interval of 10 ms, and feedback calculation is performed, so as to realize the judgment and limitation of the out-of-food weight by weighing.

[0068] Then, if the food at the outlet of the pet feeder exceeds the preset weight (for example, 500 g), or if when the preset weight is exceeded, the weight change at this moment is greater than the set single out-of-food amount (i.e., calculated by formula (6)), the out-of-food is stopped, which realizes the prevention of food accumulation caused by the pet not eating in time.

[0069] Among them, the out-of-food feedback calculation method can also be set by the following way: F_max = 100g / time, ∑F_day<500g (8); In formula (8), F_max is the single maximum out-of-food amount (for example, 100 g), and ∑F_day is the cumulative out-of-food amount in a day.

[0070] Also, when the pet feeder is out of food, i.e., the pet approaches, if at this moment the animal suddenly rushes into the target area, i.e., at this moment it is identified that there is a target animal in the target image, the out-of-food is also stopped.

[0071] Then, when the emergency stop condition is triggered, the emergency stop action is performed: the motor power is cut off to realize hardware protection; the out-of-food gate is closed; the event log (including time stamp and trigger reason) is recorded and reset. The reset includes: clearing the gesture recognition buffer, restarting the background difference algorithm, and returning to the initial state of pet detection.

[0072] Please refer to Figure 9 The embodiment of the application provides a pet feeder control system, which comprises: The detection module 10 is used for acquiring the target image in the target area in front of the pet feeder in real time.

[0073] The recognition module 20 is configured to recognize a gesture action in the target image by using a gesture recognition model when no target animal is detected in the target image.

[0074] The control module 30 is configured to control the pet feeder to execute a corresponding target instruction to control the motor of the pet feeder to start or stop when the gesture action is recognized as a target gesture.

[0075] The pet feeder control method is applicable to the pet feeder control system, and the pet feeder control system is implemented as follows: First, the image capturing module such as a camera is used to capture an image of the target area in front of the pet feeder (for example, within 10 cm), and a foreground detection formula and a pet presence determination condition are used to determine whether a pet exists in the target image.

[0076] Then, if no target animal is detected in the target image, the gesture recognition mode is entered, that is, the gesture recognition model is used to recognize the gesture action in the target area to recognize an effective gesture. For example, when a fist gesture is recognized, the grain release instruction is executed to control the motor to start to release a certain amount of grain; and when an open hand gesture is recognized, the grain release instruction is executed to control the motor to stop, thereby stopping the grain release.

[0077] It can be understood that in the present application, the target image of the target area in front of the pet feeder is detected, and when no target animal is detected, the gesture action in the target area is recognized, and when the gesture action is a target gesture, the motor of the pet feeder is controlled to start or stop to correspondingly control whether the grain is released, thereby realizing the natural gesture interaction control to automatically release the grain, improving the convenience, and realizing the intelligent and humanized pet feeding experience.

[0078] The present application also provides a terminal device, which includes a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to make the terminal device execute the pet feeder control method or the functions of the modules in the pet feeder control system.

[0079] The processor can be an integrated circuit chip with a processing capability of signals. The processor can be a general-purpose processor, including a central processing unit (CPU), a graphics processing unit (GPU), and a network processor (NP), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, a discrete gate or transistor logic device, a discrete hardware component, at least one of the above. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor or the like, which can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application.

[0080] The memory can be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM) and the like. The memory is used to store a computer program. After receiving an execution instruction, the processor can execute the computer program accordingly.

[0081] The present application also provides a computer readable storage medium for storing the computer program used in the terminal device. For example, the computer readable storage medium can include, but is not limited to, a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk and various program code storage media.

[0082] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can also be implemented by other manners. The apparatus embodiments described above are merely illustrative, for example, the flowcharts and structural diagrams in the drawings show the possible implementation architecture, function and operation of the apparatus, method and computer program product according to the embodiments of the present application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that in alternative implementation manners, the functions noted in the blocks can also occur in different order from that noted in the drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and sometimes they can also be executed in reverse order, depending on the functions involved. It should also be noted that each block in the structural diagram and / or flowchart, and the combination of blocks in the structural diagram and / or flowchart, can be implemented by a dedicated hardware-based system for executing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0083] In addition, each functional module or unit in the embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0084] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application.

[0085] The above describes only the specific implementation manners of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A pet feeder control method characterized by, The method comprises: real-time acquisition of a target image in a target area in front of a pet feeder; when no target animal is detected in the target area through the target image, gesture action recognition of the target image is performed by using a gesture recognition model; when the gesture action is identified as a target gesture, the pet feeder is controlled to execute a corresponding target instruction.

2. The pet feeder control method according to claim 1, wherein The gesture action recognition of the target area by using the gesture recognition model further comprises: when the gesture action is not recognized, or when the gesture action is not identified as the target gesture, the target image of the target area in front of the pet feeder is continuously acquired.

3. The pet feeder control method of claim 1, wherein, The real-time acquisition of the target image in the target area in front of the pet feeder further comprises: differential operation of a current target image and a plurality of historical target images, and determination of whether the current target image has changed according to an operation result; when the current target image has changed, it is determined that the target animal exists in the current target image, and the gesture action is stopped from being recognized. The plurality of historical target images are a plurality of continuous preset images before the current target image.

4. The pet feeder control method according to claim 3, wherein When the target instruction is a grain discharging instruction, the pet feeder is controlled to execute the grain discharging instruction, and the method further comprises: when a target animal is detected in the target area through the target image, or the grain at the outlet of the pet feeder exceeds a preset weight, the pet feeder is controlled to stop executing the grain discharging instruction.

5. The pet feeder control method of claim 1, wherein, The control of the pet feeder to execute the corresponding target instruction when the gesture action is identified as the target gesture comprises: when the gesture action is identified as a first target gesture, the pet feeder is controlled to execute a grain discharging instruction to release a preset amount of grain; when the gesture action is identified as a second target gesture, the pet feeder is controlled to stop executing the grain discharging instruction to stop releasing the grain; The first target gesture is a fist gesture, and the second target gesture is a flat gesture.

6. The pet feeder control method according to claim 5, wherein, The preset amount is calculated by a grain discharging amount control formula; wherein the grain discharging amount control formula is: ; wherein Q is the grain discharging amount, N is the number of motor rotations, d is the diameter of a screw propeller in the pet feeder, L is the lead of the screw, and p is the bulk density of the grain.

7. The pet feeder control method of claim 1, wherein, The gesture action recognition of the target area by using the gesture recognition model further comprises: After a plurality of gesture sample data are collected, the gesture sample data are subjected to data enhancement processing to obtain processed data; the processed data are trained by using a classification loss function and a bounding box regression loss function, and the gesture recognition model is obtained when a target number of iterations is reached.

8. A pet feeder control system characterized by, The method comprises: a detection module configured to acquire a target image in a target area in front of a pet feeder in real time; an identification module configured to, when no target animal is detected in the target area through the target image, perform gesture action recognition of the target image by using a gesture recognition model; a control module configured to, when the gesture action is identified as a target gesture, control the pet feeder to execute a corresponding target instruction, so as to control the motor of the pet feeder to be turned on or turned off.

9. A pet feeder, characterized by The method comprises: A memory, a processor, and a computer program stored on the memory and executable on the processor, which, when executed by the processor, implement the steps of the pet feeder control method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, A computer program stored on a memory, which, when executed by a processor, implements the steps of the pet feeder control method according to any one of claims 1-7.

Images