Fish tank dynamic interaction display device based on position detection and interaction implementation method

Abstract

The invention discloses a fish tank dynamic interaction display device based on position detection and an interaction implementation method. Belongs to the technical field of intelligent aquarium equipment and digital interaction and collects fish body images in real time, a control screen supports touch and voice interaction, and a background display screen dynamically updates display content according to the position of a fish body. The control processing hardware module integrates a microprocessor, a real-time digital graphic processing chip and an artificial intelligence processing unit, and image correction, multi-fish body detection tracking, coordinate mapping and dynamic background rendering are achieved. The method comprises the steps of image acquisition, geometric correction, position detection, coordinate mapping, dynamic rendering, cross-device linkage and the like, and supports gesture interaction, AI background generation and multi-fish tank linkage functions. Through combination of real-time position detection and a dynamic display technology, immersive interactive experience of fish moving and scene following is constructed, the problem that a traditional fish tank is insufficient in interactivity is solved, and the fish tank is suitable for families, commercial display and other scenes.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent aquarium equipment and digital interaction, and particularly relates to a dynamic interactive display device for a fish tank based on position detection and an interaction implementation method. Background Art

[0002] In the technical field of aquarium equipment, traditional fish tanks have relatively single functions, mainly used for fish breeding and basic viewing, and it is difficult to achieve dynamic interaction with fish. With the development of digital technology, intelligent aquarium equipment has gradually become the development trend of the industry, but the existing technologies still have significant deficiencies in terms of real-time interactivity and user experience.

[0003] Chinese Patent CN110536086A discloses an artificial intelligence liquid crystal TV fish tank. The device includes a liquid crystal TV and a control box, and the basic display function is realized through the function modules in the control box. However, it cannot dynamically adjust the display content according to the real-time position information of the fish, and the interactivity is weak. Essentially, it still belongs to the mode of "static display + preset program".

[0004] The multimedia 3D simulation fish tank system disclosed in Chinese Patent CN109840829A, although it realizes the combination of the client and the cloud server, focuses on the presentation of 3D effects, lacks a dynamic interaction mechanism with real fish, and fails to establish a real-time mapping relationship between the fish body movement trajectory and the virtual scene.

[0005] The Internet social interaction electronic fish tank system described in Chinese Patent CN104589888B realizes networking through the WiFi module, but its technical focus is on the social interaction function, and there is insufficient technical implementation in real-time visual interaction (such as the linkage between fish body position detection and dynamic display), and it cannot meet the user's demand for "real-time fish-screen interaction".

[0006] The accurate touch interaction method and system based on the MR fish tank proposed in Chinese Patent CN111897423B, although it solves the touch accuracy problem in the MR scenario, only focuses on the touch offset compensation in human-computer interaction, does not involve the real-time interaction between the fish body and the display content, and the interaction dimension is limited to the "human-device" level, lacking the dynamic association of "device-biology".

[0007] The virtual fish tank and its implementation method disclosed in Chinese Patent CN107329633A, based on laser projection and computer image generation technology, focuses on the presentation of virtual effects, but lacks an interaction mechanism with real fish. Essentially, it belongs to a pure virtual display scheme of "virtual fish tank without physical objects" and cannot meet the dynamic interaction requirements of real aquarium scenes.

[0008] The core defects of the existing technologies are mainly reflected in: 1. Visual interaction tomography: The display system lacks a real-time linkage mechanism with the dynamic state of the fish body and cannot generate corresponding dynamic effects in real time according to the position changes of the fish, resulting in a fragmented experience of "the fish moves but the scene doesn't"; for example, when the fish swims, the background screen cannot synchronously present linkage effects such as the swaying of waterweeds and the change of light and shadow, and the viewing experience remains in a fragmented state of "static fish tank + fixed background".

[0009] 2. Weak anti-interference ability of image recognition: In the fish tank environment, background light sources or reflections on the display screen are likely to generate image noise on the glass surface. Traditional cameras use general-purpose processors to analyze images (with a processing speed of only 15 - 20 frames per second), resulting in a lag in detecting the position of the fish body (the end-to-end delay can reach more than 200 ms), and the dynamic background update cannot match the real-time movement of the fish body.

[0010] 3. Single interaction method: It relies on physical buttons or simple touches, lacks natural interaction methods such as voice commands and gesture recognition, and the background content is limited to preset templates, unable to generate personalized scenes according to user preferences (such as the voice command "generate an underwater volcano scene"), and the depth of interaction is insufficient.

[0011] 4. Lack of multi-device collaboration: In commercial display or home cluster scenarios, multiple fish tanks cannot synchronize the fish population position data, and the background displays of each device are independent, making it difficult to create an immersive overall experience.

[0012] In the prior art, the development of the visual interaction function of fish tanks is still in its infancy. On the one hand, there is a lack of a real-time linkage mechanism between the display system and the dynamic state of the fish body, and it is impossible to generate and display corresponding dynamic effects in real time according to the position changes of the fish; on the other hand, the human-computer interaction method is single, mostly relying on preset programs or simple operations, making it difficult to achieve in-depth interaction between users and the display content of the fish tank. In addition, the prior art has obvious deficiencies in aspects such as the anti-interference ability of image recognition, the system response speed, and the interaction experience. Moreover, the prior art lacks support for user gesture interaction, personalized background generation, and multi-device collaborative display. For example, it is impossible to dynamically adjust the background effect according to user gestures, it is difficult to generate customized scenes based on natural language commands, and cross-device linkage of fish population movement cannot be achieved between multiple fish tanks, restricting the scalability of the interaction scenario.

[0013] Therefore, how to organically combine image recognition, dynamic display, and multi-modal interaction technologies to systematically solve the technical bottlenecks of insufficient interactivity and fixed scenarios of traditional fish tanks, provide a dynamic interaction solution for intelligent aquarium equipment, and significantly improve the user experience and product competitiveness has become an urgent technical problem in this field. Summary of the Invention

[0014] In view of this, the purpose of the present invention is to overcome the deficiencies of insufficient real-time interactivity, susceptibility to interference in image recognition, and single interaction method in the existing intelligent fish tanks, and to provide a fish tank dynamic interaction display device and an interaction implementation method based on position detection. By using a micro camera to capture the position information of fish bodies in real time, a polarization filter lens is used to shield the background reflection to improve the clarity of image acquisition. With the help of a dedicated digital graphics processing chip and a neural network processor, a millisecond-level response for fish body detection and tracking is achieved, and multiple interaction modules such as voice and gesture, as well as a cloud AI background generation function, are integrated to establish a real-time linkage mechanism between the fish body movement trajectory and the dynamic background, thus solving the disconnection problem of "the fish moves while the scene remains static" in traditional fish tanks, realizing a closed loop of "position detection - intelligent processing - dynamic display - natural interaction", and providing users with a highly immersive, highly interactive and rich-scene intelligent aquarium experience.

[0015] To achieve the above object, in the first aspect of the present invention, a fish tank dynamic interaction display device based on position detection is provided, including: The fish tank main body; A micro camera, which is used to collect fish body images in the fish tank in real time and output image signals. A polarization filter lens is provided in front of the micro camera lens to shield the light interference of the background display screen; A control screen, which is a touch screen hardware unit integrated with a microphone, and is used to display the system status and receive user instructions; A background display screen, which is installed behind the fish tank main body and is used to play a dynamic background and display a dynamic effect according to the position of the fish; A control processing hardware module, which includes a microprocessor, a real-time digital graphics processing chip, an artificial intelligence processing unit and a storage unit. The microprocessor is connected to the micro camera, the control screen and the background display screen, and is used for: Processing the image signal through the real-time digital graphics processing chip to obtain the position information of the fish; Controlling the background display content according to the user instruction; Converting the fish position information into an image control signal including dynamic effect parameters and outputting it to the background display screen. Among them, the real-time digital graphics processing chip integrates a geometric transformation unit and a mask generation unit, which are respectively used for image correction and fish tank area cropping. The artificial intelligence processing unit is used to perform multi-fish body detection, tracking and coordinate mapping. The control processing hardware module also includes a coordinate mapping unit and a display space transformation unit, which are used to map the fish body coordinates to the display coordinate system, and integrates a physics engine, which is used to calculate the dynamic background effect parameters based on the fish body position.

[0016] Furthermore, a gesture interaction module is further included. The gesture interaction module includes a gesture recognition camera and a deep neural network processor. The gesture recognition camera is used to collect gesture images, and the deep neural network processor is used to recognize gestures and trigger dynamic effects.

[0017] Further, the control processing hardware module further includes a multi-device networking interface, which supports synchronizing the fish body position data of multiple fish tanks through a local area network to achieve cross-device background linkage.

[0018] The second aspect of the present invention provides a method for realizing dynamic interaction of a fish tank based on position detection, which uses the above-mentioned device. The signal processing flow of fish body position detection and dynamic background linkage includes the following steps: S1. Collect the fish tank video stream through a micro camera with a polarization filter lens, and transmit the image data to the control processing hardware module; S2. Perform geometric correction, size normalization, elevation distortion correction, and fish tank area cropping on the image signal through a real-time digital graphics processing chip, specifically including: S21. Dynamically calculate the scaling factor based on the image width and height parameters, and compress the image to a preset size; S22. Perform image rotation transformation, perspective correction, and scaling operations through a geometric transformation unit; S23. Perform perspective distortion correction on the elevation shooting image based on the camera internal parameter matrix; S24. Generate a rectangular cropping mask through a mask generation unit to eliminate redundant backgrounds; S3. Call the model through the artificial intelligence processing unit to perform multi-fish body detection, tracking, and coordinate mapping, and generate structured data including fish body position coordinates and velocity vectors, specifically including: S31. Construct a backbone network using depthwise separable convolution and cross-stage residual structure, and output multi-scale feature maps; S32. Generate candidate boxes according to the anchor box parameters, and use a dynamic threshold strategy combined with a weighted NMS algorithm to screen target boxes; S33. Dynamically adjust the IoU threshold through the target density matrix to suppress redundant detection boxes; S34. Implement multi-target ID management through a motion-appearance dual feature association tracker; S4. Through the coordinate mapping unit and the display space transformation unit, convert the detection coordinates into a set of fish body positions and velocity vector groups in the background display coordinate system specifically including: S41. The appearance feature uses ResNet-18 to output a 128-dimensional vector, and the motion prediction uses Kalman filtering; S42. Compensate for the scaling and cropping operations through batch matrix operations, and restore the detection coordinates to the original resolution; S43. Map the set of fish body positions in the original coordinate system to the display coordinate system; S5. Dynamic rendering: According to the structured data and user instructions, calculate the dynamic effect parameters through the physics engine, generate dynamic image control signals and output them to the background display screen, specifically including: S6. After receiving the control signal, the background display screen optimizes the pixel response speed through the display driver chip, and displays the dynamic effect at the corresponding position, ensuring that the movement of the fish body is linked with the virtual background effect in real time and the visual alignment is accurate.

[0019] Furthermore, step S5 also includes: S51. Read the current background mode and fish body position / velocity data; S52. Call the GPU-accelerated hydrodynamic model to simulate the multi-ripple superposition effect, and calculate the diffusion parameter matrix based on the vertical velocity component of the fish body; S53. Calculate the fireworks particle swarm emission parameters based on the fish body velocity vector, and use spatial hashing to optimize the calculation of the overlapping area; S54. Allocate the calculation accuracy based on the fish body importance score to generate a dynamic effect with end-to-end latency; Furthermore, step S3 also includes: S35. Perform feature extraction through a depthwise separable convolutional backbone network; S36. Fuse multi-scale features through a feature pyramid network to improve the small target detection ability; S37. Dynamic threshold screening: adaptively adjust the confidence threshold according to the target density; S38. Trajectory tracking: perform associated tracking on the fish body in consecutive frames based on the Kalman filter algorithm.

[0020] Furthermore, step S5 also includes: S55. Ripple effect generation: calculate the multi-ripple superposition interference parameters based on the vertical velocity component of the fish body; S56. Particle system control: generate fireworks particle swarm emission parameters with different colors and densities according to the fish body velocity vector; S57. Physics engine simulation: calculate the physical parameters of the dynamic effect in real time through the GPU-accelerated physics engine.

[0021] Furthermore, it also includes a method for realizing the multi-aquarium linkage process, and the specific steps are as follows: S61. Real-time synchronize the fish group coordinate sets and device IDs of each aquarium through the multi-device networking interface; S62. When the fish body in the first aquarium moves to the preset area, trigger the background display screen of the second aquarium to generate a collaborative dynamic effect; S63. Build a cross-device virtual ecological scene, and synthesize the movement trajectories of the fish groups in multiple aquariums into a migration animation for segmented display. S64. Global coordinate mapping: uniformly map the fish body position data of multiple fish tanks to a shared coordinate system; S65. Linked effect rendering: generate a dynamic background effect for cross-device linkage based on the global coordinates.

[0022] Furthermore, it also includes an implementation method for the AI-generated background interface, and the specific steps are as follows: S71. Receive the user's voice command through the control screen and convert it into text by the voice recognition chip; S72. Transmit the text to the cloud StableDiffusion model through the wireless communication hardware unit; S73. Download the generated 1080p@60fps dynamic video stream and overlay it with the local fish body position data for output; S74. Identify the scene type based on the current fish body distribution and behavior pattern; S75. Send a background generation request to the cloud server through the wireless communication hardware unit; S76. Receive and cache the dynamic background materials generated by the cloud; S77. Dynamically switch the adapted background materials according to the fish body position.

[0023] Furthermore, it also includes an implementation method for gesture interaction, and the specific steps are as follows: S81. Collect the user's gesture image through the gesture recognition camera; S82. The depth neural network processor runs the MediaPipe algorithm to output the coordinates of 21 hand joints; S83. Map the gesture position (hx, hy) to the dynamic effect trigger point, and the gesture speed is positively correlated with the effect intensity; S84. Gesture feature extraction: extract gesture features through the depth neural network processor; S85. Action classification: recognize wave, click, and slide actions based on the pre-trained gesture classification model; S86. Coordinate mapping: map the gesture coordinates to the background display coordinate system; S87. Effect trigger: generate and display the corresponding dynamic effect according to the recognized gesture action and coordinate position.

[0024] The present invention adopts the above technical solutions and has the following beneficial effects: 1. Real-time dynamic linkage, enhancing the viewing interest Real-time capture the fish body position through the micro camera, combined with the control processing hardware module and the background display screen, to achieve real-time synchronous linkage between the fish body swimming and the dynamic background (such as virtual waterweed swaying, fish school interaction effect), breaking the fragmentation of the traditional fish tank with "static background + independent fish body movement", and significantly improving the real-time and interest of the viewing process.

[0025] 2. Hardware anti-interference design to improve detection accuracy The polarization filter lens in front of the micro camera lens shields the reflection of the background display screen through the principle of physical optical filtering, effectively avoiding image noise interference, improving the clarity of fish body image acquisition, significantly enhancing the position detection accuracy, and ensuring the stability and accuracy of the interaction effect.

[0026] 3. Multiple interaction methods to build a three-dimensional interaction system The capacitive touch control screen integrated with a microphone supports multiple interaction methods such as touch operation and voice commands. Users can switch the background animation and virtual scene in real time, building a three-dimensional interaction system of "user-fish-display system", breaking through the traditional single viewing mode of the fish tank, and enhancing the operation convenience and user participation.

[0027] 4. Dedicated hardware acceleration to ensure system performance The control processing hardware module adopts a dedicated image processing DSP chip and high-speed hardware interfaces such as USB3.0 and HDMI to achieve real-time transmission and processing of image signals, improving the system response speed and anti-interference ability, and providing hardware-level guarantee for the long-term stable operation of the device.

[0028] 5. Empowered by artificial intelligence to achieve personalized interaction The integrated artificial intelligence processing unit supports the simultaneous recognition and tracking of multiple fish bodies through deep learning algorithms, and adaptively adjusts the background display content according to the species, quantity and behavior patterns of the fish, providing a differentiated and personalized interaction experience.

[0029] 6. Rich extended functions to improve scene adaptability Gesture interaction function: directly trigger virtual effects (such as generating ripples by waving) through hand movements, building a more natural human-computer interaction interface and enhancing the user immersion; AI-generated background function: generate personalized dynamic scenes (such as "underwater volcano") based on voice commands, breaking through the limitations of traditional preset backgrounds and meeting the diverse needs of users; Multi-fish tank linkage system: supports cross-device fish group position data synchronization and background collaborative rendering, applicable to multiple scenarios such as home cluster display and commercial scene narration, creating a cross-device virtual ecological linkage experience.

[0030] 7. Comprehensive technological innovation to solve industry pain points Systematically integrating image recognition, dynamic display, multiple interaction and multi-device collaboration technologies, effectively solving the core problems of traditional fish tanks such as insufficient interactivity, fixed display content and weak anti-interference ability, providing a highly intelligent and highly scalable solution for intelligent aquarium equipment, with significant technological progress and market application value.

[0031] Through the innovation of the entire signal processing process of anti-interference acquisition, high-precision detection, real-time rendering, and cross-device collaboration, the present invention systematically solves the core problems of insufficient interactivity and lag in linkage of traditional fish tanks. It not only realizes the immersive experience of "the fish moves and the scene follows", but also opens up diverse application scenarios for intelligent aquarium equipment through technical modular design, with significant technological progressiveness and industrial application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0033] Figure 1 is a schematic structural diagram of the fish tank dynamic interaction display control device of the present invention Figure 1 。

[0034] Figure 2 is a schematic structural diagram of the fish tank dynamic interaction display control device of the present invention Figure 1 。

[0035] Figure 3 is a schematic diagram of the system structure of the fish tank dynamic interaction display control device of the present invention.

[0036] Figure 4 is a flowchart of the method for realizing fish tank dynamic interaction of the present invention.

[0037] In the figure: 1, fish tank main body; 2, micro camera; 3, control screen; 4, background display screen; 5, control processing hardware module; 6, polarization filter lens; 7, microprocessor; 8, real-time digital graphics processing chip; 9, storage unit; 10, wireless communication hardware unit; 11, artificial intelligence processing unit; 12, multi-device networking interface; 13, geometric transformation unit; 14, mask generation unit; 15, coordinate mapping unit; 16, display space transformation unit; 17, physics engine; 18, gesture interaction module; 19, gesture recognition camera; 20, deep neural network processor. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the invention.

[0039] Embodiment 1 Please refer to Figure 1 、Figure 2 and Figure 3 As shown in Figure 3 , this embodiment provides a fish tank dynamic interactive display device based on position detection, including: A fish tank main body 1; A micro camera 2, configured to collect fish body images in the fish tank in real time and output image signals. A polarization filter lens 6 is provided in front of the lens of the micro camera 2 to shield the light interference of the background display screen 4; A control screen 3, which is a touch screen hardware unit integrated with a microphone, and is used to display the system state and receive user instructions; A background display screen 4, installed behind the fish tank main body 1, and is used to play a dynamic background and display a dynamic effect according to the position of the fish; A control processing hardware module 5, including a microprocessor 7, a real-time digital graphics processing chip 8, an artificial intelligence processing unit 11 and a storage unit 9. The microprocessor 7 is connected to the micro camera 2, the control screen 3 and the background display screen 4, and is used for: Processing the image signal through the real-time digital graphics processing chip 8 to obtain the position information of the fish; Controlling the background display content according to the user instruction; Converting the fish position information into an image control signal including dynamic effect parameters and outputting it to the background display screen 4. Wherein, the real-time digital graphics processing chip 8 integrates a geometric transformation unit 13 and a mask generation unit 14, which are respectively used for image correction and fish tank area cutting. The artificial intelligence processing unit 11 is used to perform multi-fish body detection, tracking and coordinate mapping. The control processing hardware module 5 further includes a coordinate mapping unit 15 and a display space transformation unit 16, which are used to map the fish body coordinates to the display coordinate system, and integrates a physical engine 17, which is used to calculate the dynamic background effect parameters based on the fish body position.

[0040] As an implementation manner, this embodiment further includes a gesture interaction module 18. The gesture interaction module 18 includes a gesture recognition camera 19 and a deep neural network processor 20. The gesture recognition camera 19 is used to collect gesture images, and the deep neural network processor 20 is used to recognize gestures and trigger dynamic effects.

[0041] As an implementation manner, the control processing hardware module 5 in this embodiment further includes a multi-device networking interface 12, which supports synchronizing the fish body position data of multiple fish tanks through a local area network to realize cross-device background linkage.

[0042] Embodiment Two Please refer to Figure 1 , Figure 2 and Figure 3 As shown in Figure 1 , Figure 2 and Figure 3 , this embodiment provides a fish tank dynamic interactive display device and an interactive implementation method based on position detection, including: The fish tank main body 1 is a transparent glass container made of tempered glass with a thickness of 8 mm and a high light transmittance, and is used for raising ornamental fish; The micro camera 2 is installed in front of the base of the front of the fish tank main body 1, and includes a 1 / 2.3-inch CMOS high-resolution image sensor, supports 1080p@60fps video capture, and is connected to the fish tank base through a connecting rod with a built-in data cable, and is used for capturing the position information of the fish in real time and outputting an image signal; The polarization filter lens 6 is installed in front of the lens of the micro camera 2, and is a circular polarization optical lens with a diameter of 25 mm. It shields the background display screen 4 through the principle of physical optical filtering, and improves the anti-interference ability of fish body recognition; The control screen 3 is installed in front of the micro camera 2, and is a 7-inch capacitive touch screen hardware unit with a resolution of 1024×600 pixels, and is built with a MEMS microphone and a voice recognition chip, and is used for displaying the system status and inputting user instructions through touch or voice; The background display screen 4 is installed behind the fish tank main body 1, and is a liquid crystal display hardware module with a resolution of 1080p (1920×1080p pixels), a refresh rate of 60 Hz, and is integrated with an HDMI2.0 video receiving interface, and is used for playing a dynamic background and displaying a corresponding dynamic effect according to the position of the fish; The control processing hardware module 5 is installed behind the background display screen 4, as Figure 3 shown, and includes a microprocessor 7, a real-time digital graphics processing chip 8, a storage unit 9, a wireless communication hardware unit 10 and an artificial intelligence processing unit 11. The microprocessor 7 is connected to the micro camera 2, the control screen 3 and the background display screen 4 through a data bus, and is used for: receiving the image signal output by the micro camera 2, and obtaining the position information of the fish through the real-time digital graphics processing chip 8; according to the user instructions input by the control screen 3, controlling the display content of the background display screen 4 through a hardware interface circuit; converting the extracted fish position information into a dynamic image control signal, and outputting it to the background display screen 4 through a transmission interface.

[0043] In this embodiment, the microprocessor uses the Rockchip RK3588 chip. Its CPU is an 8-core heterogeneous architecture, including 4-core ARM Cortex-A76 (with a maximum main frequency of 2.4 GHz) and 4-core Cortex-A55 (with a maximum main frequency of 1.8 GHz), integrating an ARM Mali-G610 MP4 GPU graphics processing unit, supporting 8K video decoding and rendering. In addition, the chip is built-in with a neural network processor (NPU) with a computing power of 6 TOPS, which can implement intelligent tasks such as multi-target recognition and deep learning inference. In this embodiment, the real-time digital graphics processing chip 8 is a dedicated image processing DSP chip, integrating a convolutional neural network acceleration unit, supporting real-time target detection algorithms, and capable of processing 1080p resolution image data at a speed of 60 frames per second, realizing millisecond-level fish body position detection and tracking.

[0044] In this embodiment, the storage unit 9 includes 8GB DDR4 RAM and 128GB eMMC flash memory, which are used to store system software, dynamic background material library and user configuration data.

[0045] In this embodiment, the wireless communication hardware unit 10 is a dual-band Wi-Fi 6 chip, supporting 2.4 GHz and 5 GHz frequency bands, and is connected to the microprocessor 7 through a UART serial interface to realize device networking and remote control functions.

[0046] In this embodiment, the artificial intelligence processing unit 11 is a neural network processor (NPU) with a computing power of 6 TOPS (6 trillion operations per second), which is used to execute deep learning algorithms, realize simultaneous recognition and tracking of multiple fish bodies, and adaptively adjust the background display content according to the species, quantity and behavior patterns of the fish.

[0047] In this embodiment, the micro camera 2 is connected to the control processing hardware module 5 through a USB3.0 data interface to realize the hardware real-time transmission of fish body image signals.

[0048] In this embodiment, the control screen 3 is a capacitive touch hardware panel, supporting 10-point touch, built-in with a MEMS microphone and a voice recognition chip, and is connected to the microprocessor 7 through an I2S audio interface to realize the hardware signal conversion of touch instructions and voice instructions.

[0049] In this embodiment, the dynamic background of the background display screen 4 includes art pictures, animation effects or virtual scenes pre-stored in the storage unit 9, and are selected and loaded onto the background display screen 4 through the hardware buttons or touch interface of the control screen 3.

[0050] Embodiment Three Please combine Figure 1 , Figure 2 , Figure 3 andFigure 4 As shown in Figure 4 , this embodiment provides a method for implementing dynamic interaction of an aquarium based on position detection. Using the above-mentioned device, the signal processing flow of fish body position detection and dynamic background linkage includes the following steps: S1. Collect the aquarium video stream through the micro camera 2 with a polarization filter lens 6, and transmit the image data to the control and processing hardware module 5; S2. Perform geometric correction, size standardization, elevation distortion correction, and aquarium area cropping on the image signal through the real-time digital graphics processing chip 8, specifically including: S21. Dynamically calculate the scaling factor based on the image width and height parameters, and compress the image to a preset size; S22. Perform image rotation transformation, perspective correction, and scaling operations through the geometric transformation unit 13; S23. Perform perspective distortion correction on the elevation shooting image based on the camera internal parameter matrix; S24. Generate a rectangular cropping mask through the mask generation unit 14 to eliminate redundant background; S3. Call the model through the artificial intelligence processing unit 11 to perform multi-fish body detection, tracking, and coordinate mapping, and generate structured data including fish body position coordinates and velocity vectors, specifically including: S31. Construct a backbone network using depthwise separable convolution and cross-stage residual structure to output multi-scale feature maps; S32. Generate candidate boxes according to the anchor box parameters, and use the dynamic threshold strategy combined with the weighted NMS algorithm to screen the target boxes; S33. Dynamically adjust the IoU threshold through the target density matrix to suppress redundant detection boxes; S34. Implement multi-target ID management through a motion-appearance dual-feature association tracker; S4. Through the coordinate mapping unit (15) and the display space transformation unit (16), convert the detection coordinates into a set of fish body positions and velocity vector groups in the background display coordinate system, specifically including: S41. The appearance feature uses ResNet-18 to output a 128-dimensional vector, and the motion prediction uses the Kalman filter; S42. Compensate for the scaling and cropping operations through batch matrix operations, and restore the detection coordinates to the original resolution; S43. Map the set of fish body positions in the original coordinate system to the display coordinate system; S44. Map the set of fish body positions in the original coordinate system to the display coordinate system; S5. Dynamic rendering: According to the structured data and user instructions, calculate the dynamic effect parameters through the physics engine 17, generate a dynamic image control signal, and output it to the background display screen 4, specifically including: S51. Read the current background mode and fish body position / velocity data; S52. Call the fluid dynamics model accelerated by GPU to simulate the multi-ripple superposition effect, and calculate the diffusion parameter matrix based on the vertical velocity component of the fish body; S53. Calculate the emission parameters of the fireworks particle swarm based on the fish body velocity vector, and use spatial hashing to optimize the calculation of the overlapping area; S54. Allocate the calculation accuracy based on the fish body importance score to generate the dynamic effect of the end-to-end delay; S6. After the background display screen 4 receives the control signal, optimize the pixel response speed through the display driver chip, and display the dynamic effect at the corresponding position to ensure that the movement of the fish body is linked with the virtual background effect in real time and the visual alignment is accurate.

[0051] As a preferred implementation manner, step S3 in this embodiment further includes: S31. Perform feature extraction through a depthwise separable convolutional backbone network; S32. Fuse multi-scale features through a feature pyramid network to improve the small target detection ability; S33. Dynamic threshold screening: adaptively adjust the confidence threshold according to the target density; S34. Trajectory tracking: perform associative tracking on the fish body in continuous frames based on the Kalman filtering algorithm.

[0052] As a preferred implementation manner, step S5 in this embodiment further includes: S51. Ripple effect generation: calculate the multi-ripple superposition interference parameters based on the vertical velocity component of the fish body; S52. Particle system control: generate the emission parameters of the fireworks particle swarm with different colors and densities according to the fish body velocity vector; S53. Physical engine simulation: real-time calculate the physical parameters of the dynamic effect through the physical engine accelerated by GPU.

[0053] As a preferred implementation manner, this embodiment further includes a method for realizing the multi-aquarium linkage process, and the specific steps are as follows: S61. Real-time synchronize the fish group coordinate sets and device IDs of each aquarium through the multi-device networking interface 12; S62. When the fish body in the first aquarium moves to the preset area, trigger the background display screen of the second aquarium to generate a collaborative dynamic effect; S63. Build a cross-device virtual ecological scene, and synthesize the movement trajectories of the fish groups in multiple aquariums into migration animations for segmented display; S64. Global coordinate mapping: uniformly map the fish body position data of multiple aquariums to the shared coordinate system; S65. Linkage effect rendering: generate a cross-device linkage dynamic background effect based on the global coordinates.

[0054] As a preferred embodiment, this embodiment further includes an implementation method for the AI-generated background interface. The specific steps are as follows: S71. Receive the user's voice command through the control screen 3 and convert it into text via the voice recognition chip. S72. Transmit the text to the cloud StableDiffusion model through the wireless communication hardware unit 10; the StableDiffusion model is a generative model based on deep learning, mainly used for image generation tasks.

[0055] S73. Download the generated 1080p@60fps dynamic video stream and overlay it with the local fish body position data for output. S74. Identify the scene type based on the current fish body distribution and behavior pattern. S75. Send a background generation request to the cloud server through the wireless communication hardware unit. S76. Receive and cache the dynamic background materials generated by the cloud. S77. Dynamically switch the adapted background materials according to the fish body position.

[0056] As a preferred embodiment, this embodiment further includes an implementation method for gesture interaction. The specific steps are as follows: S81. Collect the user's gesture image through the gesture recognition camera 19. S82. The deep neural network processor 20 runs the MediaPipe algorithm to output the coordinates of 21 hand joints. S83. Map the gesture position (hx, hy) to the dynamic effect trigger point, and the gesture speed is positively correlated with the effect intensity. S84. Gesture feature extraction: Extract gesture features through the deep neural network processor. S85. Action classification: Recognize wave, click, and slide actions based on the pre-trained gesture classification model. S86. Coordinate mapping: Map the gesture coordinates to the background display coordinate system. S87. Effect trigger: Generate and display the corresponding dynamic effects according to the recognized gesture actions and coordinate positions.

[0057] Embodiment 4 Please refer to Figure 1 、 Figure 2 、 Figure 3 and Figure 4 As shown, the signal processing flow for the fish body position detection and dynamic background linkage in this embodiment includes the following steps: Step S1: The micro camera 2 uses a global shutter CMOS sensor to capture real-time images of the fish tank at a resolution of 1080p@60fps, and transmits the image data to the control processing hardware module 5 via a USB3.0 high-speed data interface.

[0058] Step S2: The real-time digital graphics processing chip 8 performs geometric transformation and feature extraction on the input video frame, specifically including: S21: Image orientation correction and size standardization The original image data is read through the video interface to obtain the image width and height parameters. A rotation transformation is used to adjust the image orientation to a standard direction. The scaling factor is dynamically calculated based on the original image size, compressing the image to a preset maximum size range to ensure that the image data volume adapts to the computing power requirements of the subsequent artificial intelligence processing unit.

[0059] S22: Upward Angle Shooting Correction and Aquarium Area Cropping For scenarios where the camera is installed below the front of the fish tank and requires shooting at an upward angle, image correction is achieved through the following operations: first, the original image shot at an upward angle is corrected for perspective distortion based on the camera's intrinsic parameter matrix and distortion coefficient to eliminate the stretching deformation of the tank edge caused by the upward tilt of the lens; then, based on the actual size and installation position of the fish tank, the fish tank boundary is manually or automatically calibrated in the corrected image, and a rectangular cropping area mask corresponding to the actual tank is generated. The redundant background outside the mask is eliminated to ensure that the extracted image area accurately matches the actual fish tank space.

[0060] Step S3: The AI processing unit 11 calls the optimized YOLOv5 deep learning model and performs the following operations on the NPU of RK3588: S31: Multi-scale feature extraction and candidate box generation The preprocessed video frames are fed into an improved YOLOv5 network (optimized for the RK3588 NPU). The backbone network utilizes depthwise separable convolutions (3×3 kernels) and a cross-stage residual structure, outputting multi-scale feature maps ranging from 13×13 to 52×52. A feature pyramid network (FPP) is used to fuse features from different levels, improving the detection of small objects in dense fish schools. Based on K-means clustering analysis (5,000 training samples) and fish aspect ratio statistics, anchor box parameters are set to [1.2, 3.5] and [2.8, 6.9], respectively. Initial candidate boxes are generated at the three detection layers, enabling simultaneous detection of more than 50 fish targets.

[0061] S32: Dynamic threshold detection optimization "Perform classification (Sigmoid) and regression (CIoULoss) predictions on the candidate bounding boxes to generate object confidence scores and localization parameters. Adopt a two-stage screening strategy: first, dynamically adjust the confidence threshold based on the object density (0.6 for low density / 0.8 for high density). The formula for the object density matrix is: Object density matrix = Number of detection boxes / Image area. The number of detection boxes is the total number of fish body target bounding boxes recognized in the current frame, and the image area is the pixel value of the effective detection region after preprocessing (unit: pixel²). Then apply the improved weighted NMS algorithm (IoU threshold dynamically adjusted from 0.5 to 0.7), and adaptively suppress redundant bounding boxes by calculating the detection box density matrix to ensure the effective separation of overlapping fish bodies. In typical scenarios, a detection recall rate of >95% can be maintained." S33: Multi-object tracking and identity preservation Construct a tracker based on the association of motion and appearance dual features: 1) For appearance features, use lightweight ResNet-18 (outputting a 128-dimensional vector, quantized by NPU INT8) to calculate the cosine similarity; 2) For motion prediction, use Kalman filtering (state vector [x, y, w, h, vx, vy]) combined with Mahalanobis distance metric; 3) Introduce an exponential decay mechanism for ID management (decay factor λ = 0.05). When the target is lost, retain the ID information for 30 frames (corresponding to 1 second @ 30fps), and achieve continuous identity after occlusion through the trajectory interpolation algorithm. Experiments show that the ID switching rate < 0.1 times / minute."

[0062] S34: Multi-resolution coordinate mapping and display space alignment For the set of multi-fish target bounding boxes finally confirmed, perform the following operations: First, restore the detection coordinates of each fish body from the resolution after AI processing (such as 800×600) to the original image resolution (such as 1080p), and compensate for the scaling and cropping operations in the preprocessing stage through batch matrix operations to generate a set of multi-fish positions in the original image coordinate system ; Then, based on the physical parameters of the display device (screen size, pixel density, DPI), construct a coordinate transformation matrix to map the original image coordinates to the display device coordinate system, generating a set of multi-fish positions in the display coordinate system , where represents the normalized coordinates of the i-th fish in the display space; at the same time, combine the position change amount between two consecutive frames to calculate the moving speed vector group of each fish body ; Finally, bind the position coordinate group, speed vector group, and fish body ID into a structured data block, and synchronize it to the physical engine in the S4 stage through the shared memory interface to achieve real-time synchronous linkage between the real fish school movement and the virtual effect."

[0063] Step S4: Calculation of multi-fish dynamic background effect parameters Step S4: Based on each group of fish body display coordinates {(px1, py1), (px2, py2),...} and their movement speed vectors {(vx1, vy1), (vx2, vy2),...} output in stage S34, the microprocessor 7 combines the currently activated background display mode (such as ripples, fireworks, etc.) and calls the GPU-accelerated physics engine to parallelly calculate the dynamic effect parameters, specifically including: S41: Multi-fish data aggregation and mode judgment The microprocessor 7 obtains the real-time position coordinate group and speed vector group of all recognized fish bodies from the multi-target tracking module, and at the same time reads the currently activated background display mode from the display control module to provide basic data for subsequent batch physical simulation calculations.

[0064] S42: Batch parameter calculation of the physics engine Perform corresponding calculations according to the current background mode: Ripple mode: Weight the vertical speed component of each fish to generate a ripple diffusion parameter matrix centered on the position of each fish, and simulate the superposition and interference effects of multiple ripples through the GPU-accelerated hydrodynamic model. Fireworks mode: Calculate independent fireworks emission parameters based on the speed vector of each fish to generate multiple groups of particle swarms with different colors, densities, and lifecycles, and achieve the synchronous blooming effect of multiple fireworks through GPU parallel computing.

[0065] S43: Multi-target performance optimization Implement measures such as batch processing of fish school parameters through the GPU parallel computing architecture, spatial hashing optimization of physical effects in overlapping areas, and dynamic adjustment of calculation precision allocation based on the importance score of fish bodies.

[0066] Step S5: Signal conversion and high-speed output Step S5: The microprocessor 7 integrates the calculated dynamic effect parameters with the original fish school video stream, converts them into a dynamic image control signal in HDMI2.0 format, encodes them in YUV420 format through a dedicated video encoder, and uses the HDMI2.0 high-speed video interface to achieve signal output of 1080p@60fps, ensuring the synchronous transmission of the video stream and dynamic effect parameters and a smooth and non-stuttering picture.

[0067] Step S6: The background display screen 4 uses a 60Hz high refresh rate panel. After receiving the control signal, it optimizes the pixel response speed through the display driver chip, displays the dynamic effect at the corresponding position, and controls the delay from signal reception to complete display update to ≤12ms, ensuring the real-time linkage and accurate visual alignment of fish body movement and virtual background effects (such as ripples, fireworks).

[0068] The entire signal processing process uses a hardware acceleration pipeline (such as a dedicated image processor for parallel processing of acquisition, calculation, and encoding), multi-threaded software optimization (such as synchronous execution of detection and rendering tasks), and a timing calibration mechanism (such as a frame synchronization phase-locked loop technology) to control the end-to-end latency from image acquisition to display update within 50 milliseconds, ensuring a smooth and natural interaction between the fish body and the background effect perceived by the user without obvious lag.

[0069] As an extended implementation, the device can achieve the following extended functions: 1. Gesture interaction process: The environmental camera (miniature camera 3) captures user gesture images at a resolution of 640×480 and a frame rate of 30fps, and transmits them to the control and processing hardware module through the USB3.0 interface: The real-time digital graphics processing chip 8 runs the MediaPipe gesture recognition algorithm, outputs the coordinates of 21 hand joints, and generates the gesture position ((hx, hy)) in the display coordinate system after coordinate mapping; The microprocessor 7 calls the corresponding background effect function according to the gesture type (such as waving, making a fist), for example: The waving action triggers a rippling effect, with the center of the ripple being ((hx, hy)), and the diffusion speed being positively correlated with the amplitude of the gesture swing.

[0070] AI-generated background process: The user says "I want a tropical rainforest background" through the control screen 3, and the built-in voice recognition chip converts the voice into the text "tropical rainforest background"; The microprocessor 7 sends the text to the cloud AI service through the Wi-Fi module, and the cloud returns a generated 1080p@60fps dynamic video stream; The video stream is transmitted to the background display screen 4 through the HDMI interface and superimposed with the dynamic effects (such as ripples, fireworks) generated by the real-time fish school positions for display.

[0071] Multi-aquarium linkage process: Aquarium A (device ID: 001) and Aquarium B (device ID: 002) are bound through the cloud platform, and their respective main cameras upload the fish school coordinates to the server in real time; When the fish body in Aquarium A moves to the right side of the screen, the coordinate data is synchronized to Aquarium B through the cloud; The microprocessor 7 in Aquarium B generates a water flow animation in the corresponding direction on the left side of the background display screen according to the received coordinates, creating a visual effect of "fish school migration" across aquariums.

[0072] The working principle and application scenario of the present invention: Users can select preset scenarios (such as "Underwater World", "Starry Prairie") through the touch interface of the control screen 3, or trigger effects through voice commands (such as "Activate the interactive mode"). The control processing hardware module 5 retrieves the dynamic background material library in the storage unit 9 and synchronously updates the content of the background display screen 4.

[0073] As a home intelligent decoration, the present invention can be installed in the living room or study. Parents can set an interactive mode of "fish chasing virtual bait" for children through the control screen 3 - when the fish swim towards the bait icon, the background display screen 4 synchronously shows the effect of splashing water, enhancing the fun of parent - child interaction; at night, it can be switched to the "Quiet Starry Sky" mode, and the background generates a meteor animation with gradually changing starlight matching the swimming trajectory of the fish body, creating an immersive viewing experience.

[0074] In commercial application scenarios, such as aquariums, pet stores or exhibition halls, the background display screen 4 of the present invention can play virtual ecological scenes (such as coral reefs, tropical rainforests). When the fish swim, the virtual creatures (such as turtles, jellyfish) in the background will make avoidance or following actions, forming a dynamic symbiotic visual effect to attract viewers to stop. Merchants can also remotely update the background content of all display fish tanks through the wireless communication hardware unit 10 (Wi - Fi chip) of the control processing hardware module 5 to uniformly manage the display effect.

[0075] In an office environment, the present invention can be used as an intelligent stress - relieving device. The artificial intelligence processing unit 11 analyzes the behavior patterns of the fish and automatically adjusts the background display content to create a visual effect that conforms to the current environmental atmosphere, enhancing the beauty and comfort of the office space.

[0076] The present invention also supports connecting to the smart home system through the wireless communication hardware unit 10 to achieve linkage with other smart devices. For example, it can automatically adjust the brightness parameter of the background display screen 4 according to the indoor lighting brightness; or automatically switch to the "silent mode" during a specific time period (such as meeting time), turning off the dynamic effects and only retaining the basic viewing function.

[0077] Although the embodiments of the invention have been shown and described above, it can be understood that the above - mentioned embodiments are exemplary and should not be construed as limitations on the invention. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above - mentioned embodiments within the scope of the invention.

Claims

1. A fish tank dynamic interaction display device based on position detection, characterized in that, Comprising: A fish tank main body (1); A micro camera (2) for collecting fish body images in the fish tank in real time and outputting image signals. A polarization filter lens (6) is provided in front of the lens of the micro camera (2) to shield the light interference of the background display screen (4); A control screen (3), which is a touch screen hardware unit integrated with a microphone, for displaying system status and receiving user instructions; A background display screen (4), installed behind the fish tank main body (1), for playing a dynamic background and displaying a dynamic effect according to the position of the fish; A control processing hardware module (5), which includes a microprocessor (7), a real-time digital graphics processing chip (8), an artificial intelligence processing unit (11) and a storage unit (9). The microprocessor (7) is connected to the micro camera (2), the control screen (3) and the background display screen (4) and is used for: Processing the image signal through the real-time digital graphics processing chip (8) to obtain the position information of the fish; Controlling the background display content according to the user instruction; Converting the fish position information into an image control signal including dynamic effect parameters and outputting it to the background display screen (4). Among them, the real-time digital graphics processing chip (8) integrates a geometric transformation unit (13) and a mask generation unit (14), which are respectively used for image correction and fish tank area cropping. The artificial intelligence processing unit (11) is used for performing multi-fish body detection, tracking and coordinate mapping. The control processing hardware module (5) also includes a coordinate mapping unit (15) and a display space transformation unit (16) for mapping the fish body coordinates to the display coordinate system, and integrates a physical engine (17) for calculating dynamic background effect parameters based on the fish body position.

2. The device according to claim 1, characterized in that It further includes a gesture interaction module (18). The gesture interaction module (18) includes a gesture recognition camera (19) and a deep neural network processor (20). The gesture recognition camera (19) is used for collecting gesture images, and the deep neural network processor (20) is used for recognizing gestures and triggering dynamic effects.

3. The device according to claim 1, characterized in that The control processing hardware module (5) further includes a multi-device networking interface (12), which supports synchronizing the fish body position data of multiple fish tanks through a local area network to achieve cross-device background linkage.

4. A method for realizing dynamic interaction of an aquarium based on position detection, characterized in that: Using the device according to any one of claims 1 to 3, the signal processing flow of fish body position detection and dynamic background linkage includes the following steps: S1. Collect the fish tank video stream through the micro camera (2) with a polarization filter lens (6) and transmit the image data to the control processing hardware module (5); S2. Perform geometric correction, size standardization, elevation distortion correction and fish tank area cropping on the image signal through the real-time digital graphics processing chip (8), specifically including: S21. Dynamically calculate the scaling factor based on the image width and height parameters and compress the image to a preset size; S22. Perform image rotation transformation, perspective correction and scaling operations through the geometric transformation unit (13); S23. Perform perspective distortion correction on the elevation shooting picture based on the camera internal parameter matrix; S24. Generate a rectangular cropping mask through the mask generation unit (14) to eliminate redundant background; S3. The artificial intelligence processing unit (11) is called to execute multi-fish body detection, tracking, and coordinate mapping by using a model, and structured data including fish body position coordinates and velocity vectors is generated. Specifically, it includes: S31. A backbone network is constructed by using depthwise separable convolution and cross-stage residual structure to output multi-scale feature maps. S32. Candidate boxes are generated according to anchor box parameters, and a dynamic threshold strategy combined with a weighted NMS algorithm is used to filter target boxes. S33. The IoU threshold is dynamically adjusted through a target density matrix to suppress redundant detection boxes. S34. Multi-object ID management is realized through a motion-appearance dual-feature correlation tracker. S4. Through the coordinate mapping unit (15) and the display space transformation unit (16), Convert the detected coordinates into the set of fish body positions in the background display coordinate system and the velocity vector group Specifically, it includes: S41. The appearance feature uses ResNet-18 to output a 128-dimensional vector, and the motion prediction uses Kalman filtering. S42. The detection coordinates are restored to the original resolution by compensating for the scaling and cropping operations through batch matrix operations. S43. The set of fish body positions in the original coordinate system is mapped to the display coordinate system. S5. Dynamic rendering: According to the structured data and user instructions, the physical engine (17) calculates dynamic effect parameters, generates a dynamic image control signal, and outputs it to the background display screen (4). S6. After receiving the control signal, the background display screen (4) optimizes the pixel response speed through the display driver chip, and displays the dynamic effect at the corresponding position, ensuring that the fish body movement and the virtual background effect are linked in real time and the visual alignment is accurate.

5. The method according to claim 4, wherein: Specifically included in step S5 is: S51. Read the current background mode and fish body position / velocity data. S52. Call the GPU-accelerated hydrodynamic model to simulate the multi-ripple superposition effect, and calculate the diffusion parameter matrix based on the vertical velocity component of the fish body. S53. Calculate the fireworks particle swarm emission parameters based on the fish body velocity vector, and use spatial hashing to optimize the calculation of the overlapping area. S54. Allocate calculation precision based on the fish body importance score to generate a dynamic effect with end-to-end delay.

6. The method according to claim 4, wherein: Also included in step S3 is: S35. Feature extraction is performed through a depthwise separable convolution backbone network. S36. The multi-scale features are fused through a feature pyramid network to improve the small target detection ability. S37. Dynamic threshold screening: The confidence threshold is adaptively adjusted according to the target density. S38. Trajectory tracking: The fish body in continuous frames is associated and tracked based on the Kalman filter algorithm.

7. The method according to claim 6, wherein: Also included in step S5 is: S55. Ripple effect generation: Calculate the multi-ripple superposition interference parameters based on the vertical velocity component of the fish body. S56. Particle system control: Generate fireworks particle swarm emission parameters with different colors and densities according to the fish body velocity vector. S57. Physical engine simulation: The physical parameters of the dynamic effect are calculated in real time through the GPU-accelerated physical engine.

8. The method according to any one of claims 4 to 7, characterized in that: Also included is a method for realizing the multi-aquarium linkage process. The specific steps are as follows: S61. The fish group coordinate sets and device IDs of each aquarium are synchronized in real time through a multi-device networking interface. S62. When the fish body in the first aquarium moves to a preset area, the background display screen of the second aquarium is triggered to generate a collaborative dynamic effect. S63. Build a cross-device virtual ecological scenario, synthesize the movement trajectories of fish groups in multiple fish tanks into a migration animation and display it in segments; S64. Global coordinate mapping: Uniformly map the fish body position data of multiple fish tanks to a shared coordinate system; S65. Linkage effect rendering: Generate a dynamic background effect for cross-device linkage based on the global coordinates.

9. The method according to any one of claims 4 to 7, characterized in that: It also includes an implementation method for the AI-generated background interface, and the specific steps are as follows: S71. Receive the user's voice command through the control screen (3) and convert it into text by the voice recognition chip; S72. Transmit the text to the cloud through the wireless communication unit (10); S73. Download the generated dynamic video stream and superimpose it with the local fish body position data for output; S74. Identify the scene type based on the current fish body distribution and behavior pattern; S75. Send a background generation request to the cloud server through the wireless communication unit; S76. Receive and cache the dynamic background materials generated by the cloud; S77. Dynamically switch the adapted background materials according to the fish body position.

10. The method according to any one of claims 4 to 7, characterized in that: It also includes an implementation method for gesture interaction, and the specific steps are as follows: S81. Collect the user's gesture image through the gesture recognition camera; S82. The depth neural network processor outputs the hand joint coordinates; S83. Map the gesture position (hx, hy) to a dynamic effect trigger point, and the gesture speed is positively correlated with the effect intensity; S84. Gesture feature extraction: Extract gesture features through the depth neural network processor; S85. Action classification: Recognize wave, click, and slide actions based on the pre-trained gesture classification model; S86. Coordinate mapping: Map the gesture coordinates to the background display coordinate system; S87. Effect trigger: Generate and display the corresponding dynamic effect according to the recognized gesture action and coordinate position.

Images