Gesture recognition light control system

Abstract

The utility model relates to a gesture recognition light control system, which comprises a high-definition camera, an embedded artificial intelligence experiment box and a lamp panel suite, the high-definition camera and the lamp panel suite are respectively connected with the embedded artificial intelligence experiment box; the embedded artificial intelligence experiment box comprises a power supply, an MIPS processor, a neural network acceleration rod, a radiator, a mouse, a keyboard and a display, wherein the MIPS processor is electrically connected with the power supply, the neural network acceleration rod, the radiator, the mouse, the keyboard and the display. Gesture images are collected through a high-definition camera, gesture recognition is carried out through a trained neural network model in an embedded artificial intelligence experiment box, gestures can be accurately recognized, the lamp panel suite is controlled to be turned on, and real-time control is achieved. According to the utility model, image acquisition and image identification are completed through local equipment, calculation can be carried out without the help of a network or a cloud server, the privacy of a user can be better protected, and the intelligent use experience of the user is greatly improved.

Description

Technical Field

[0001] This utility model relates to the field of artificial intelligence control technology, specifically a gesture recognition lighting control system. Background Technology

[0002] Traditional lighting control methods mainly rely on mechanical switches, remote controls, or mobile apps. These methods have significant limitations: mechanical switches require users to operate them manually at close range, and frequent contact can cause wear and tear on the buttons, and they cannot achieve coordinated control of lights in multiple areas; remote control can achieve operation at a certain distance, but it has problems such as numerous buttons that are prone to accidental touches and the need to be pointed at the receiver; mobile app control requires users to take out their phones, unlock them, and open the corresponding program, which is cumbersome and cannot meet the needs of real-time control.

[0003] With the development of artificial intelligence interaction technology, gesture recognition technology is gradually being applied to the field of lighting control. Current lighting control methods have significant shortcomings in terms of ease of operation, environmental adaptability, and functionality. In particular, image acquisition and recognition are performed via networks or cloud servers, which fails to adequately protect user privacy and cannot meet users' needs for intelligent and user-friendly control.

[0004] Therefore, developing a non-contact, environmentally adaptable, personalized gesture control system that prevents user privacy leaks and can be linked with other devices has become an urgent need to address the shortcomings of existing technologies and improve user experience. Utility Model Content

[0005] To address the problems existing in the prior art, the purpose of this utility model is to provide a gesture recognition lighting control system that can accurately recognize gestures to control lighting without compromising user privacy.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A gesture recognition lighting control system includes a high-definition camera, an embedded artificial intelligence experimental box, and a light board kit;

[0008] The embedded artificial intelligence experimental box includes a MIPS processor, which stores a gesture recognition light control instruction program containing a trained neural network model.

[0009] The high-definition camera is connected to the MIPS processor to capture gesture images and send them to the MIPS processor;

[0010] The MIPS processor is connected to the light panel kit and is used to process gesture images through a trained neural network model to obtain gesture recognition results, and then convert the gesture recognition results into gesture signals that control the on / off state of the light panel kit.

[0011] Furthermore, the embedded artificial intelligence experimental kit also includes a neural network accelerator stick, which is connected to the MIPS processor to accelerate the MIPS processor's recognition of gesture images.

[0012] Furthermore, the embedded artificial intelligence experimental box also includes a power supply, heat sink, mouse, keyboard and monitor, which are respectively connected to the MIPS processor;

[0013] The power supply is a 5V / 3A power supply, used to convert 220V AC power to 5V DC power to power the MIPS processor;

[0014] The heatsink is used to dissipate heat from the MIPS processor;

[0015] The mouse and keyboard are used to input gesture recognition lighting control commands into the MIPS processor;

[0016] The display is used to show the processing results of the MIPS processor.

[0017] Furthermore, the high-definition camera and light board kit are connected to the MIPS processor via USB interfaces.

[0018] Furthermore, the MIPS processor has the following built-in features:

[0019] Power interface, for electrical connection to the power supply;

[0020] HDMI interface, for electrical connection to the monitor;

[0021] The USB 3.0 interface is electrically connected to the neural network accelerator stick and the high-definition camera.

[0022] USB 2.0 interface, for electrical connection to mouse, keyboard and light board kit;

[0023] It has a 3-pin connector for electrical connection to the heatsink.

[0024] Furthermore, the lamp board kit includes, in sequence, an LED controller, multiple sets of lamp boards and lamp board connectors, and a 24V / 1.5A power adapter kit. The power adapter kit has a built-in power connector, which is electrically connected to the lamp board. The LED controller is connected to the MIPS processor.

[0025] Furthermore, the LED controller is equipped with an STM32 controller, a controller connector, and a Type-C interface. The STM32 controller is electrically connected to the controller connector and the Type-C interface, respectively. The Type-C interface of the LED controller is electrically connected to the USB driver circuit of the embedded artificial intelligence experimental box.

[0026] Furthermore, the MIPS processor is a MIPS processor that runs the Linux system.

[0027] Furthermore, the neural network accelerator stick supports LoongArch, X86, ARM and MIPS platforms, and has a built-in CNN network acceleration engine with a data throughput of ≥2GB / s.

[0028] Furthermore, the system can be scaled to train different neural network models to support complex gesture control.

[0029] In summary, this utility model has the following advantages:

[0030] This invention enables contactless control of lights based on gesture recognition using a high-definition camera, significantly enhancing the user's intelligent experience. The gesture control is unaffected by ambient light, offering higher accuracy and stronger performance. Furthermore, the system utilizes a neural network accelerator in an embedded AI experimental kit for accelerated processing, resulting in stable recognition quality, low computational load, and fast processing speed. The neural network model can be trained on other platforms to develop more complex gesture-controlled light models. This system boasts strong scalability and flexibility, enabling it to handle not only complex gesture controls but also adapt flexibly to different hardware environments, user needs, and application scenarios, creating an intelligent and personalized interactive experience. Image acquisition and recognition are both performed locally, eliminating the need for network or cloud server computation, thus better protecting user privacy and improving security and user experience. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the structure of a gesture recognition lighting control system according to the present invention.

[0032] Figure 2 This is a schematic diagram of the lamp panel kit structure.

[0033] Figure 3 This is a diagram of the first gesture.

[0034] Figure 4 This is a diagram of the second gesture.

[0035] Figure 5 This is a diagram of the third gesture.

[0036] Figure 6 This is a diagram of the fourth gesture.

[0037] Figure 7 This is a diagram of the fifth gesture.

[0038] Figure 8 This is a diagram of the sixth gesture.

[0039] Figure 9 This is a diagram of the seventh gesture.

[0040] Explanation of reference numerals in the attached figures:

[0041] 1. LED controller; 2. First LED board; 3. First LED board connecting piece; 4. Second LED board; 5. Second LED board connecting piece; 6. Third LED board; 7. Third LED board connecting piece; 8. Fourth LED board; 9. Fourth LED board connecting piece; 10. Fifth LED board; 11. Power adapter kit; 12. STM32 controller; 13. Controller connecting piece; 14. Type-C interface; 15. Power connector. Detailed Implementation

[0042] The purpose of this invention is to provide a gesture recognition lighting control system that can accurately recognize gestures to control lighting, has good stability, scalability and flexibility, and can protect user privacy.

[0043] The present invention will now be described in further detail.

[0044] The implementation steps of the technical solution of this utility model include:

[0045] Step 101: The embedded artificial intelligence experimental box runs a gesture recognition Linux program, which drives the high-definition camera to turn on.

[0046] Step 102: The user makes a gesture, the high-definition camera captures the gesture image, and the gesture data is sent to the MIPS processor in the embedded artificial intelligence experimental box via the USB 3.0 interface.

[0047] Step 103: The MIPS processor receives gesture information, preprocesses the gesture image, and extracts the feature vector of the gesture contour features.

[0048] Step 104: The neural network accelerator obtains the gesture recognition result through the trained neural network model and sends it to the MIPS processor.

[0049] Step 105: The MIPS processor converts the gesture recognition result into a gesture signal, and finally controls the light in the light panel kit to turn on or off through the gesture signal. Real-time control of the light can be achieved by changing the gesture.

[0050] The gesture recognition result, when the user gesture is recognized as Figure 3The first gesture shown is raising the index finger, which illuminates the first light panel 2 in the light panel kit.

[0051] The gesture recognition result, when the user gesture is recognized as Figure 4 The second gesture shown – raising the index and middle fingers – illuminates the first light panel 2 and the second light panel 4 in the light panel kit.

[0052] The gesture recognition result, when the user gesture is recognized as Figure 5 The third gesture shown – raising the index, middle, and ring fingers – illuminates the first light panel 2, the second light panel 4, and the third light panel 6 in the light panel kit.

[0053] The gesture recognition result, when the user gesture is recognized as Figure 6 The fourth gesture shown – raising the index, middle, ring, and little fingers – illuminates the first light panel 2, the second light panel 4, the third light panel 6, and the fourth light panel 8 in the light panel kit.

[0054] The gesture recognition result, when the user gesture is recognized as Figure 7 The fifth gesture shown – raising all five fingers – illuminates the first light panel 2, the second light panel 4, the third light panel 6, the fourth light panel 8, and the fifth light panel 10 in the light panel kit.

[0055] The gesture recognition result, when the user gesture is recognized as Figure 8 The sixth gesture shown is a clenched fist. In this case, all light panels in the light panel kit, including the first light panel 2, the second light panel 4, the third light panel 6, the fourth light panel 8, and the fifth light panel 10, will be turned off.

[0056] The gesture recognition result, when the user gesture is recognized as Figure 9 The seventh gesture shown is the OK gesture. In this case, all light panels in the light panel kit, including the first light panel 2, the second light panel 4, the third light panel 6, the fourth light panel 8, and the fifth light panel 10, will be lit.

[0057] This invention utilizes existing artificial intelligence image recognition technology to determine different gestures by analyzing the relative positions of key joints in the gestures, thereby controlling the corresponding functions.

[0058] like Figure 1 As shown, this utility model provides a gesture recognition lighting control system, including a high-definition camera, an embedded artificial intelligence experimental box, and a light board kit.

[0059] A high-definition camera is used to capture gesture images, and the gesture data is sent to the MIPS processor in the embedded artificial intelligence experimental box via a USB 3.0 interface.

[0060] The embedded artificial intelligence experimental box includes a power supply, a MIPS processor, a neural network accelerator, a heat sink, a mouse, a keyboard, and a monitor. The MIPS processor is electrically connected to the power supply, the neural network accelerator, the heat sink, the mouse, the keyboard, and the monitor.

[0061] The power supply is required to be 5V / 3A. The power supply can convert 220V AC power into 5V DC power to power the MIPS processor in the embedded artificial intelligence experimental box, providing stable voltage and power.

[0062] The MIPS processor, a CPU architecture, specifically the Loongson 2K1000 processor, is designed for industrial control and terminal applications. It utilizes the GS264 or GS464 processor core, equipped with a 32KB L1 instruction cache, a 32KB L1 data cache, and a shared 1MB L2 cache. It supports various storage and display device connections and has a typical power consumption of 3.5W. This MIPS processor features a built-in power interface for electrical connection to a power supply; an HDMI interface for electrical connection to a monitor; a USB 3.0 interface for electrical connection to neural network accelerators and high-definition cameras; a USB 2.0 interface for electrical connection to mice, keyboards, and LED light panel kits; and a 3-pin interface for electrical connection to a heatsink.

[0063] The Neural Network Accelerator Stick is a high-performance, general-purpose deep learning accelerator stick. It supports mainstream platforms such as LoongArch, x86, ARM, and MIPS. It features a built-in CNN network acceleration engine, enabling high-performance, low-power acceleration of CNN network models. Its advanced TPU architecture design efficiently completes face detection, tracking, feature extraction, and recognition of multiple dynamic video streams, and effectively supports attribute detection such as sunglasses, masks, gender, and age. In this invention, the Neural Network Accelerator Stick obtains gesture recognition results through a trained neural network model and sends them to the MIPS processor. It is electrically connected to the MIPS processor via a USB 3.0 interface, with a data throughput ≥2GB / s.

[0064] The heatsink is used to dissipate heat from the MIPS processor in a timely manner. It starts working when the power is turned on and is electrically connected to the MIPS processor via a 3-pin connector.

[0065] The mouse and keyboard are used to input Linux commands and are electrically connected to the MIPS processor via a USB 2.0 interface.

[0066] The monitor displays the Linux command input process in real time and brings up a camera dialog box to display gesture images in real time. It is electrically connected to the MIPS processor via an HDMI interface.

[0067] It should be emphasized that, for those skilled in the art, Figure 1The schematic diagram shown is merely an example of a preferred embodiment; those skilled in the art can explore further. Figure 1 The schematic diagram of the gesture recognition lighting control system shown can be easily supplemented with new structures; the names of each structure are custom names and are only used to understand the various structures of this system, and are not used to limit the technical solution of this utility model. The core of the technical solution of this utility model is the function to be achieved by each custom structure.

[0068] like Figure 2 As shown, this utility model provides a lamp board kit, including an LED controller 1, a first lamp board 2, a first lamp board connecting piece 3, a second lamp board 4, a second lamp board connecting piece 5, a third lamp board 6, a third lamp board connecting piece 7, a fourth lamp board 8, a fourth lamp board connector 9, a fifth lamp board 10, and a 24V / 1.5A power adapter kit 11.

[0069] The LED controller 1 includes an STM32 controller 12; a controller connector 13; and a Type-C interface 14 for controlling the lighting and extinguishing of the first LED board 2, the second LED board 4, the third LED board 6, the fourth LED board 8, and the fifth LED board 10.

[0070] The 24V / 1.5A power adapter kit 11 has a built-in power connector 15 that is electrically connected to the fifth lamp board 10. The power connector 15 adopts a gold-finger plug-in electrical connection structure.

[0071] The MIPS processor converts the gesture recognition result into an electrical signal, which is transmitted to the Type-C interface 14 via the USB 3.0 interface. The Type-C interface 14 is electrically connected to the STM32 controller 12, energizing the controller connector 13. The controller connector 13 uses a gold-finger pluggable electrical connection structure and is connected to the first LED board 2. The first LED board connector 3 uses a gold-finger pluggable electrical connection structure and is connected to the first LED board 2 and the second LED board 4. The second LED board connector 5 uses a gold-finger pluggable electrical connection structure and is connected to the second LED board 4 and the third LED board 6. The third LED board connector 7 uses a gold-finger pluggable electrical connection structure and is connected to the third LED board 6 and the fourth LED board 8. The fourth LED board connector 9 uses a gold-finger pluggable electrical connection structure and is connected to the fourth LED board 8 and the fifth LED board 10.

[0072] The gesture recognition result, when the user gesture is recognized as Figure 3 As shown, when the controller connector 13 is energized, the first lamp board 2 lights up.

[0073] The gesture recognition result, when the user gesture is recognized as Figure 4 As shown, when the controller connecting piece 13 and the first lamp board connecting piece 3 are energized, the first lamp board 2 and the second lamp board 4 light up.

[0074] The gesture recognition result, when the user gesture is recognized as Figure 5 As shown, when the controller connecting piece 13, the first lamp board connecting piece 3, and the second lamp board connecting piece 5 are energized, the first lamp board 2, the second lamp board 4, and the third lamp board 6 light up.

[0075] The gesture recognition result, when the user gesture is recognized as Figure 6 As shown, when the controller connecting piece 13, the first lamp board connecting piece 3, the second lamp board connecting piece 5, and the third lamp board connecting piece 7 are energized, the first lamp board 2, the second lamp board 4, the third lamp board 6, and the fourth lamp board 8 light up.

[0076] The gesture recognition result, when the user gesture is recognized as Figure 7 As shown, when the controller connecting piece 13, the first lamp board connecting piece 3, the second lamp board connecting piece 5, the third lamp board connecting piece 7, and the fourth lamp board connecting piece 9 are powered, the first lamp board 2, the second lamp board 4, the third lamp board 6, the fourth lamp board 8, and the fifth lamp board 10 light up.

[0077] The gesture recognition result, when the user gesture is recognized as Figure 8 As shown, the action of clenching five fingers into a fist corresponds to a high-level pulse width of ≤100ms in the control signal, triggering all light boards to power off. The controller connection piece 13, the first light board connection piece 3, the second light board connection piece 5, the third light board connection piece 7, and the fourth light board connection piece 9 lose power, and the first light board 2, the second light board 4, the third light board 6, the fourth light board 8, and the fifth light board 10 all go out.

[0078] The gesture recognition result, when the user gesture is recognized as Figure 9 As shown, when the controller connecting piece 13, the first lamp board connecting piece 3, the second lamp board connecting piece 5, the third lamp board connecting piece 7, and the fourth lamp board connecting piece 9 are powered, the first lamp board 2, the second lamp board 4, the third lamp board 6, the fourth lamp board 8, and the fifth lamp board 10 are all lit.

[0079] In summary, this invention provides a gesture recognition lighting control system. When a user makes a gesture, a high-definition camera captures the gesture image. The image is then processed by a trained neural network model in an embedded artificial intelligence experimental box for gesture recognition. This accurately identifies the gesture and controls the lighting of the light panel kit, achieving real-time control with high gesture recognition accuracy. Image acquisition and recognition are both performed locally, eliminating the need for network or cloud server computation, thus better protecting user privacy and significantly enhancing the user's intelligent experience. This patent utilizes a neural network accelerator to improve image recognition speed; the measured recognition time delay is ≤50ms, while existing solutions generally have a delay >100ms.

[0080] The above embodiments are preferred embodiments of the present utility model, but the embodiments of the present utility model are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present utility model shall be considered equivalent substitutions and shall be included within the protection scope of the present utility model.

Claims

1. A gesture recognition lighting control system, characterized in that: It includes a high-definition camera, an embedded artificial intelligence experimental box, and a light board kit; The embedded artificial intelligence experimental box includes a MIPS processor. The MIPS processor stores a gesture recognition lighting control instruction program, and the instruction program contains a trained neural network model; The high-definition camera is connected to the MIPS processor and is used to collect gesture images and send them to the MIPS processor; The MIPS processor is connected to the light board kit and is used to process the gesture images through the trained neural network model to obtain gesture recognition results, and convert the gesture recognition results into gesture signals for controlling the lighting on and off states of the light board kit.

2. The lighting control system according to claim 1, wherein: The embedded artificial intelligence experimental box further includes a neural network acceleration stick. The neural network acceleration stick is connected to the MIPS processor and is used to accelerate the MIPS processor's recognition of gesture images.

3. The lighting control system according to claim 2, wherein: The embedded artificial intelligence experimental box further includes a power supply, a radiator, a mouse, a keyboard, and a display respectively connected to the MIPS processor; The power supply is a 5V / 3A power supply and is used to convert 220V alternating current into 5V direct current to supply power to the MIPS processor; The radiator is used to dissipate heat from the MIPS processor; The mouse and keyboard are used to input the gesture recognition lighting control instruction program into the MIPS processor; The display is used to display the processing results of the MIPS processor.

4. The lighting control system according to claim 3, wherein: The high-definition camera and the light board kit are respectively connected to the MIPS processor through USB interfaces.

5. The lighting control system according to claim 4, wherein: The MIPS processor is built-in with: A power interface, electrically connected to the power supply; An HDMI interface, electrically connected to the display; A USB3.0 interface, electrically connected to the neural network acceleration stick and the high-definition camera; A USB2.0 interface, electrically connected to the mouse, keyboard, and light board kit; A 3pin interface, electrically connected to the radiator.

6. The lighting control system according to claim 1, wherein: The light board kit includes a lamp bead controller, multiple groups of light boards and light board connecting pieces, and a 24V / 1.5A power adapter set connected in sequence. The power adapter set is built-in with a power connection piece, and the power connection piece is electrically connected to the light board; The lamp bead controller is connected to the MIPS processor.

7. The lighting control system according to claim 6, wherein: The lamp bead controller is configured with a STM32 controller, a controller connecting piece, and a Type-C interface. The STM32 controller is electrically connected to the controller connecting piece and the Type-C interface respectively; the Type-C interface of the lamp bead controller is electrically connected to the USB drive circuit of the embedded artificial intelligence experimental box.

8. The lighting control system according to claim 1, wherein: The MIPS processor is a MIPS processor running the Linux system.

9. The lighting control system according to claim 1, wherein: The neural network acceleration stick supports LoongArch, X86, ARM, and MIPS platforms, and is built-in with a CNN network acceleration engine, and the data throughput ≥ 2GB / s.

10. The lighting control system according to claim 1, wherein: The system can be extended to train different neural network models to support complex gesture control.

Images