Feeding device and method for monitoring the amount of food in a feeding device

Abstract

The application discloses a kind of feeding device residual food monitoring method and intelligent bird feeder system based on visual identification.The method comprises: obtaining the image containing bin transparent window area collected by imaging device;Image segmentation is carried out to identify food filling area and idle area, and the equivalent material level average height of surface inclined plane is calculated using multi-point sampling weighted compensation algorithm, and then the current residual food inventory parameter is obtained and fed back to client.The application extracts time-stamped time series inventory data, establishes a moving average consumption model, and realizes dynamic prediction of the remaining feeding time.The system reduces hardware costs by reusing bird watching cameras (combined with wide-angle distortion correction or mirror surface), and performs intelligent early warning for food leakage or blockage by comparing real-time consumption rate with historical average rate.In summary, the application breaks through the limitations of traditional sensors, realizes quantitative whole process, and has high robustness and predictability for residual food intelligent management.

Description

Technical Field

[0001] This invention relates to the field of smart home and wildlife monitoring equipment technology, specifically to a method for monitoring leftover food in a feeder based on visual recognition and a smart bird feeder system. Background Technology

[0002] With increasing awareness of natural ecological protection, smart bird feeders have become an important tool for observing and protecting wild birds. To ensure a continuous food supply for birds, timely monitoring and early warning of remaining food have become core requirements for smart feeders.

[0003] In existing technologies, such as the Chinese utility model patent CN210580461U, a feeder with residual feed detection function is disclosed, which senses the presence of feed by fixing an infrared photoelectric switch to the inner wall of the feed hopper. However, in practical applications, such solutions have significant limitations: 1. Insufficient monitoring accuracy: The infrared beam detection system can only provide binary feedback of "grain present" or "grain absent", which is a breakpoint detection. It cannot know the specific percentage or volume change of grain in the silo in real time, so users cannot grasp the dynamic process of grain consumption. 2. Poor environmental interference resistance: The silo environment is enclosed and dusty, and the surface of the infrared detector is easily blocked by grain dust or grain sticking to the wall, resulting in false alarms and low reliability. 3. Lack of predictive support: Existing detection methods cannot predict the remaining feeding time based on current consumption trends, making it difficult for users to plan refeeding times in advance according to their travel or work schedules; 4. Hardware redundancy and power consumption conflict: To achieve high-precision monitoring, a large number of sensors need to be deployed, which not only increases hardware costs and assembly difficulty, but also leads to a significant increase in the standby power consumption of battery-powered devices.

[0004] Therefore, there is an urgent need in this field for an intelligent surplus grain monitoring solution that can reuse existing hardware resources, has high-precision quantification capabilities, and can provide trend prediction. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention aims to provide a technical solution that can solve the above problems.

[0006] This invention provides a visual recognition-based method for monitoring leftover food in a bird feeder and an intelligent bird feeder system. Through the deep integration of intelligent visual algorithms and physical structural logic, it achieves accurate monitoring of leftover food and dynamic trend prediction.

[0007] (I) Surplus Grain Monitoring Method Based on Visual Recognition The method provided by this invention is implemented through the following detailed steps: 1. Image acquisition including the transparent window area of ​​the silo This solution utilizes the imaging device built into the feeder to perform the image acquisition task. To balance monitoring accuracy and system power consumption, the image acquisition process includes any of the following triggering methods or a combination thereof: Event triggering logic: The system utilizes pyroelectric infrared (PIR) sensors or visual AI detection algorithms to monitor the feeder's trough or standing area in real time. When a bird trigger signal is detected, the system simultaneously activates the camera to capture birdwatching footage and simultaneously captures an image frame containing the transparent window area of ​​the feed trough as the image to be processed. This achieves "zero additional power consumption" synchronization between remaining feed monitoring and birdwatching activities.

[0008] Clock compensation logic: The system has a preset sampling period (e.g., 6 hours, 12 hours, or 24 hours). If no bird-triggered event occurs within the sampling period, the system will forcibly wake up the imaging device to acquire a sample image to ensure the continuity of the remaining grain time series data.

[0009] 2. Image Recognition and Region Segmentation The system preprocesses the acquired raw image, including median filtering for noise reduction, histogram equalization for contrast enhancement, and lens distortion correction. Subsequently, the algorithm locates transparent window regions using a pre-defined mask or feature point matching.

[0010] By using edge detection operators (such as the Canny operator) and texture analysis, the system identifies two distinct regions within a transparent window: Food-filled areas: Because food (such as grains, seeds, etc.) has a specific color depth (usually dark) and rough texture characteristics, it appears as a high-frequency information area in the image; Idle area: The empty part inside the hopper corresponding to the transparent window, usually presents a uniform background color or high reflectivity, and is a low-frequency information area.

[0011] 3. Accurate calculation of material level height (slope compensation algorithm) Considering that the top surface of the grain in the silo is not perfectly flat due to gravity and the position of the discharge port, it often presents an irregularly stacked sloping surface. Single-point measurement would lead to serious errors. Therefore, this invention adopts a multi-point sampling weighted compensation algorithm.

[0012] The system uniformly extracts n sampling points along the boundary between the identified filled and empty areas, where the pixel ordinate of each sampling point i is... Let the pixel ordinate of the bottom baseline of the image be... The conversion constant between pixels and physical size is k.

[0013] The formula for calculating the physical material level height at each sampling point is: To obtain the most representative stock level, the system introduces a weighting coefficient. Calculate the equivalent average material level height : in, The value is preset based on the weight of the sampling point's horizontal position in the image. This is achieved through calculation. Total height of transparent window The proportion is used to derive the current surplus grain inventory parameters (such as percentage Φ or volume value). ).

[0014] 4. Dynamic trend modeling and remaining time prediction The system models data by recording time-stamped sequence data. A timestamp refers to the absolute time T at which the imaging device captured the sample image.

[0015] Define real-time consumption rate .set up The remaining grain inventory was measured at all times. , The remaining grain inventory was measured at all times. ,but: To smooth out the instantaneous fluctuations in consumption caused by occasional gatherings of bird flocks, the system uses a moving average model to calculate the average consumption rate. : Where m is the number of historical sampling periods selected. Finally, the predicted remaining feeding time is... The calculation is as follows: The predicted value is dynamically fed back to the user's client, enabling the user to plan the timing of food replenishment in advance.

[0016] 5. Logic stress testing under abnormal conditions The system calculates the current consumption rate in real time. With average consumption rate Deviation value δ: Grain leakage risk diagnosis: If (in If the sensitivity threshold is preset (e.g., 50%), it is determined that there is damage to the silo or mechanical failure causing abnormal food loss, and an alarm is triggered. Food jam risk diagnosis: If the system records a large number of bird-triggered signals, but the material level remains unchanged for a long period of time (i.e., If the condition is not met, it is determined that the food is unable to be fed normally due to dampness and clumping or blockage by foreign objects, and the user is prompted to check the hopper.

[0017] (II) Structure of the Intelligent Bird Feeder System The present invention also provides a hardware system for implementing the above method, comprising: 1. Shell Structure: It is equipped with a hopper for storing feed, and the side wall of the hopper has a transparent window. The bottom of the hopper has a guide ramp that slopes towards the discharge port. The angle of the ramp matches the natural angle of repose of the feed to optimize the accuracy of material position identification.

[0018] 2. Imaging device layout: Wide-angle multiplexing mode: The imaging device is mounted on the front of the housing and uses a wide-angle lens. Its mounting position ensures that the edge of the lens's imaging area can cover the transparent window on the side, and the window image is acquired using a distortion correction algorithm.

[0019] Auxiliary camera mode: A miniature auxiliary camera specifically designed for the transparent window is located inside the housing or on the side.

[0020] Reflected light path mode: A reflective mirror is installed on the underside of the eaves. The main axis of the imaging device is pointed towards the feeding area, and real-time images of the transparent window of the side hopper are acquired through the optical refraction path of the reflective mirror.

[0021] 3. Control processor: integrates image processing and communication modules to execute the complex calculation models described above and interact with the cloud or APP.

[0022] Compared with the prior art, the beneficial effects of the present invention are: 1. Comprehensiveness of quantitative monitoring: It breaks through the limitations of traditional sensors that are "point- or non-point-based" and realizes linear quantitative monitoring of the entire process from full warehouse to empty warehouse, providing users with accurate "surplus grain inventory parameters"; 2. Intelligent predictive capabilities: By establishing a consumption model with timestamps, it has achieved a leap from "state monitoring" to "trend prediction" for the first time, greatly improving the user experience; 3. Ingenious hardware design: By using wide-angle multiplexing or a reflective mirror solution, "one machine serves two purposes" is achieved, obtaining extremely high technological added value without increasing or with minimal increase in hardware costs; 4. Robustness of the algorithm: The introduction of a multi-point sampling weighted compensation algorithm and a moving average rate model effectively overcomes the measurement errors caused by uneven grain stacking and fluctuations in bird feeding. 5. Proactive equipment maintenance: By utilizing abnormal rate diagnostic logic, mechanical or environmental faults such as food jams and leaks can be proactively detected, ensuring the continuity and safety of feeding activities for wild birds.

[0023] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the overall system structure and optical path of the intelligent bird feeder of the present invention; Figure 2 This is a flowchart of the visual recognition-based grain surplus monitoring method of the present invention; Figure 3 This is the logic diagram of dynamic consumption modeling and remaining feeding time prediction of the present invention; Figure 4 This is the abnormal consumption diagnosis and system triggering logic diagram of the present invention; Figure 5 This is a schematic diagram of the structure of a bird feeder, one of the wide-angle edge reuse solutions of the present invention; Figure 6 This is a schematic diagram of the structure of the food tray and imaging device of the present invention; Figure 7 This is a schematic diagram of the structure of one type of bird feeder according to the reflective scheme of the present invention; Figure 8 This is a schematic diagram of the transparent window and imaging device of the present invention.

[0026] The reference numerals and names in the figure are as follows: 10 Shell; 11 Roof; 12 Solar charging panel; 13 Reflective mirror; 20 Hopper; 21 Guide ramp; 22 Discharge port; 23 Transparent window; 30 Feeding tray; 40 Imaging device. Detailed Implementation

[0027] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figures 1 to 8In this embodiment of the invention, a method for monitoring leftover food in a feeder based on visual recognition and an intelligent bird feeder system are provided. Through the deep integration of intelligent visual algorithms and physical structural logic, the system achieves accurate monitoring of leftover food and prediction of dynamic trends.

[0029] I. Example 1: Hardware Integration and Optical Path Layout of an Intelligent Bird Feeder In this embodiment, a smart bird feeder hardware system is provided. The bird feeder includes a main housing 10, a rainproof roof 11 located above the housing 10, a feeding dish 30 located at the bottom, and a feed hopper 20 disposed inside the housing 10. The side wall of the feed hopper 20 is provided with at least one transparent window 23, which is made of transparent acrylic or glass, so that an external imaging device 40 can observe the real-time height of the food inside the feed hopper 20.

[0030] To achieve grain surplus detection without increasing hardware costs, this embodiment focuses on optimizing the optical path between the imaging device 40 and the transparent window 23, specifically including the following three optional solutions: 1. Wide-angle edge multiplexing scheme: The imaging device 40 uses a main camera with a wide-angle field of view of 110 to 160 degrees, mounted on the front of the feeder (facing the feeding dish 30). By precisely adjusting the position of the transparent window 23 of the feed hopper 20, it is positioned in the edge distortion area of ​​the main camera's image. After acquiring the original image, the control processor performs distortion correction and local cropping on the edge area using a preset algorithm, thereby extracting a clear local image of the transparent window 23. This method achieves "dual-purpose functionality," enabling both high-definition recording of birds feeding and monitoring of remaining food.

[0031] 2. Reflective Reflection Solution: A reflective mirror 13 with a specific tilt angle is installed on the underside of the eaves of the roof 11. This reflective mirror 13 is configured to reflect the image of the transparent window 23 of the hopper 20, located on the front or side of the housing 10, to the center of the field of view of the main camera mounted on the front. This solution resolves the contradiction that the hopper 20 window and the camera are on the same side and cannot directly image, and the reflective mirror 13 is hidden under the eaves, making it less susceptible to rain or dust erosion.

[0032] 3. Multi-camera collaboration solution: In the high-performance version, in addition to the main camera, the system is also equipped with a low-power, low-pixel miniature auxiliary camera. This auxiliary camera is installed close to the outside of the transparent window 23 (for example, installed at the underside of the eaves directly opposite the transparent window 23) and is specifically used to acquire images of the grain. It is connected to the control processor via an internal bus with the main camera.

[0033] 4. Hardware Implementation Environment Description: To implement the aforementioned method for monitoring surplus grain, this invention provides a preferred hardware implementation platform. For example... Figures 5 to 8 As shown, the bird feeder includes a housing 10 with a feed bin 20 and a feeding dish 30. The spatial layout of the imaging device 40, in coordination with the physical structure of the feed bin 20, provides a high-standard data source for visual analysis. ① Optical sensing closed loop: The imaging device 40 and the transparent window 23 on the side wall of the hopper 20 are configured with an enclosing angle of less than 180 degrees (centripetal layout), or a reflection path is formed by a convex reflector set at the bottom of the eaves of the roof 11. This structure ensures that the edge field of view or reflected field of view of the imaging device 40 can completely cover the transparent window 23 without adding an additional camera, and the surface of the transparent window 23 is provided with a height reference mark to provide a quantitative reference for the conversion of pixel height and physical margin.

[0034] ② Physical steady-state compensation: The bottom of the hopper 20 is equipped with a guide ramp 21 (preferably with an inclination angle of 30°-45°) adapted to the natural angle of repose of the grain. This ramp not only ensures that the grain slides smoothly to the discharge port 22, but more importantly, it guides the top grain to form a relatively flat and regular slope, thereby making the "height line" captured by the visual recognition algorithm highly repeatable and accurate, eliminating the volume calculation error caused by the random accumulation of grain.

[0035] ③ Interaction and Protection Support: To ensure the algorithm operates in all weather conditions, the device also integrates an infrared sensor wake-up module, a supplementary light, and a sound-emitting device. The infrared sensor is responsible for triggering the low-power wake-up logic, the supplementary light ensures color reproduction for nighttime recognition, and the sound-emitting device executes corresponding actions such as food shortage alarms, bird attractants, or bird scare actions based on the abnormal recognition results output by the algorithm.

[0036] II. Example 2: Material Position Recognition and Sloping Surface Compensation Algorithm Based on Image Recognition After obtaining a partial image of the transparent window, this embodiment uses the following steps to calculate the accurate residual grain parameters: 1. Region segmentation and feature extraction Because the silo walls at the transparent window are light-colored (white or gray), while food (such as grains, seeds, and sunflower seeds) are typically dark-colored (brown, black, or yellow), the control processor identifies the top surface boundary of the food accumulation by comparing two regions with significant color differences in the local image: the upper empty area and the lower food area. Using a grayscale thresholding algorithm, the processor identifies this boundary.

[0037] 2. Horizontal Compensation and Multi-Point Sampling Logic Considering that the grain will flow towards the discharge port under gravity, resulting in a non-horizontal slope on the top surface, this embodiment incorporates a guide ramp at the bottom of the hopper, sloping towards the discharge port, to improve recognition accuracy. The shape of the slope formed by the top grain during its descent matches the bottom ramp.

[0038] The algorithm uniformly selects n (e.g., n≥5) sampling points along the identified boundary line. The system obtains the vertical pixel coordinates of each sampling point. And convert it into physical height according to the preset resolution constant. To obtain the most accurate equivalent height, a weighting coefficient is introduced. The calculation formula is: Sampling points closer to the discharge port are given higher weight to reflect the true fluid dynamic residue. Ultimately, through... The remaining grain inventory parameter (e.g., 60% remaining or 500 ml remaining) is calculated based on the ratio of the total height of the transparent window.

[0039] III. Example 3: Dynamic Consumption Modeling and Multi-level Early Warning Interaction Strategy This embodiment describes how to transform static image data into intelligent predictive information.

[0040] 1. Timestamp-based time series modeling The control processor assigns an absolute timestamp to each valid grain inventory check record. For example, the system records that the inventory is 80% at time T1 and 75% at time T2.

[0041] The system establishes a moving average model, and calculates the average consumption rate by analyzing the food consumption curve over the past 48-72 hours. (Unit: % / day or ml / day). Based on this rate, the system automatically predicts the remaining feeding time (number of days).

[0042] 2. User-defined and intelligent reminder logic The system will push the predicted remaining feeding days to the user's mobile app. Users can customize the reminder threshold based on their own busy schedule. Regular reminder: If the user sets "remind 3 days in advance", a regular push will be sent when the system predicts that there are ≤3 days left to feed the baby.

[0043] Urgent Reminder: If the system detects that the rate of food consumption suddenly exceeds 150% of the average rate, it will be judged as "food leakage anomaly"; if the system detects frequent bird activity but the food height does not change in 24 hours, it will be judged as "food jam anomaly", and the system will immediately trigger a pop-up alarm.

[0044] Near-term reminder strategy: When there are less than 24 hours left, the reminder frequency will automatically increase to once every 3 hours, but will remain silent during the preset "do not disturb" period (such as 22:00 to 07:00 the next day), only message push will be retained and no sound or light alarm will be issued.

[0045] IV. Example 4: Ultra-low power operation mode To extend the feeder's battery life, this embodiment employs an "on-demand data acquisition" power management strategy: The camera is normally in sleep mode. The system uses a point-of-infrared (PIR) sensor to monitor the feeding area. When a bird flies in and triggers the PIR, the system wakes up the camera to take a photo or short video of the bird, and simultaneously captures an image of the transparent window to calculate remaining food. This "incidental capture" mechanism avoids the energy waste of separately waking up the camera to detect remaining food. If no birds visit throughout the day, the system will be forcibly woken up once a day at a fixed time (such as 10:00 AM or 4:00 PM) for supplementary measurement.

[0046] V. Example 5: Auxiliary parts, material specifications, and external operating conditions 1. Material and optical enhancement details Transparent window material: The transparent window is preferably made of UV-resistant polycarbonate (PC) or high-transparency acrylic (PMMA) to prevent yellowing caused by long-term outdoor exposure, which could affect image recognition accuracy. Physical scale lines or frosted texture areas can be preset on the inner or outer surface of the transparent window. The physical scale lines serve as a height reference scale in the algorithm, assisting the control processor in automatically calibrating the conversion constant k between pixels and actual height.

[0047] 2. Auxiliary alarm components interact with local systems. Audible and visual alarm module: In addition to push notifications from the APP, the feeder housing integrates a buzzer or miniature speaker and multi-color LED indicator lights. When the remaining feed level is below a preset extremely low threshold (e.g., 5%), the local buzzer will emit a short alarm, or the LED indicator light will change from solid green to flashing red, directly reminding maintenance personnel near the device to add feed.

[0048] 3. Physical compensation logic of hardware structure Discharge port and guide ramp: such as Figure 2 and Figure 4 As shown, the slope angle θ at the bottom of the silo is designed to be between 30° and 50° (preferably 45°). This angle design references the natural angle of repose of commonly used grains, ensuring that the grains spontaneously gather towards the discharge port under gravity. This structure corresponds to the "slope compensation logic" in the algorithm, making the resulting downward slope regular, thereby reducing the computational load of the algorithm when dealing with irregular accumulations.

[0049] 4. External Conditions and Energy Management Network connectivity environment: This system operates in a wireless communication network environment (such as Wi-Fi, 4G / 5G, or Bluetooth Low Energy (BLE)). The control processor periodically uploads the calculated remaining feed data (remaining feed inventory parameters or predicted remaining feeding time) to the cloud server via the built-in communication module, and then distributes it to the user's smart terminal device.

[0050] Energy supply solution: Considering the convenience of outdoor installation, the feeder is equipped with a high-capacity lithium battery pack and a solar charging panel 12. The control processor supports a "low-power sleep-event wake-up" mechanism. When there are no birds triggering the event and the scheduled monitoring point has not been reached, both the camera device and the high-speed processor are in a powered-off or ultra-low-power mode.

[0051] 5. Algorithm initialization and calibration process After the initial installation of the equipment or a change of the hopper model, the user can enter "calibration mode" via a mobile app. At this time, the imaging device acquires a baseline image showing the hopper completely empty. The user then manually or automatically labels the boundary pixel coordinates of the transparent window. and The initialization parameters are stored in non-volatile memory and serve as the basis for all subsequent calculations.

[0052] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

Claims

1. A method for monitoring uneaten food in a feeder based on visual recognition, characterized in that, Includes the following steps: The image to be processed, including the transparent window area of ​​the feed hopper, is acquired by the imaging device of the feeder; Identify the food-filled areas and empty areas within the transparent window region of the image to be processed; The current remaining grain inventory parameters are calculated based on the characteristic parameters of the grain filling area and the idle area. The current grain inventory parameters are fed back to the client.

2. The method for monitoring surplus grain according to claim 1, characterized in that, The steps for acquiring the image to be processed include any of the following methods or a combination thereof: The imaging device is triggered to take pictures at preset time intervals; The imaging device is triggered to take a picture when a moving object is detected within a preset range by a sensor. The image to be processed is acquired simultaneously when the imaging device detects bird feeding behavior.

3. The method for monitoring surplus grain according to claim 1, characterized in that, The steps for calculating the current remaining grain inventory parameters include: sampling multiple points in the transparent window area to obtain multiple height coordinate values ​​of the grain accumulation surface; calculating the equivalent grain height line based on the multiple height coordinate values; and calculating the current remaining grain inventory parameters by combining the total height of the transparent window.

4. The method for monitoring surplus grain according to claim 1, characterized in that, The method also includes a surplus grain trend prediction step: acquiring multiple surplus grain inventory parameter records at different time points, and establishing a food consumption model in combination with timestamps; calculating the food consumption rate, and predicting the remaining feeding time based on the current surplus grain inventory parameters and the food consumption rate.

5. The method for monitoring surplus grain according to claim 4, characterized in that, The step of providing feedback to the client also includes: receiving a user-defined warning threshold; and pushing a feeding reminder to the client when the remaining feeding time reaches the warning threshold.

6. The method for monitoring surplus grain according to claim 4, characterized in that, The method also includes an abnormal state detection step: comparing the real-time calculated food consumption rate with the historical average consumption rate; if the deviation between the real-time consumption rate and the historical average consumption rate exceeds a preset range, an abnormal warning is sent to the client.

7. An intelligent bird feeder system, characterized in that, include: A housing, wherein a hopper is provided inside the housing, and a transparent window is provided on the side wall of the hopper; An imaging device, mounted on the housing, is used to acquire an image to be processed that includes the transparent window area; A control processor, connected to the imaging device, is used to execute the grain surplus monitoring method as described in any one of claims 1-6.

8. The intelligent bird feeder system according to claim 7, characterized in that, The optical path between the imaging device and the transparent window is achieved in the following way: The imaging device is a camera with a wide-angle lens, and the transparent window is located at the edge of the imaging area of ​​the wide-angle lens; or, The imaging device includes a main camera and an auxiliary camera, the auxiliary camera being positioned facing the transparent window; or... The housing is provided with a reflective mirror, and the imaging device acquires the image of the transparent window through the reflective mirror.

9. The intelligent bird feeder system according to claim 7, characterized in that: The transparent window has a preset contrast difference in image features with the non-transparent part of the shell, which assists the control processor in performing region segmentation.

10. The intelligent bird feeder system according to claim 7, characterized in that: The system further includes a storage medium on which a computer program is stored, which, when executed by the control processor, implements the steps of the method according to any one of claims 1-6.

Images