Alarm method based on AI video analysis algorithm and intelligent terminal equipment

Abstract

The invention provides an alarm method based on an AI video analysis algorithm and intelligent terminal equipment. The method comprises the following steps: acquiring media data sent by at least one piece of interconnection equipment; analyzing the media data in real time through an AI video analysis algorithm; when an abnormal condition is detected, master control alarm information is generated; the abnormal condition is used for indicating that the user has dangerous behaviors or potential safety hazards occur in the environment. According to the technical scheme provided by the embodiment of the invention, real-time analysis of the media data can be efficiently completed through an AI video analysis algorithm. According to the identified different abnormal conditions, different types of master control alarm information are generated in time, the risk source can be accurately identified, the intelligent degree of alarm is improved, and the gas first-aid repair efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of AI video analysis technology, and in particular to an alarm method and intelligent terminal device based on an AI video analysis algorithm. Background Art

[0002] Gas is an important energy source in our lives. Gas facilities include gas pipelines, valves, pressure regulators, etc. In daily use, the gas system may encounter a series of emergency situations such as failures, which require rapid response and emergency repairs to ensure the safety of personnel and the normal operation of equipment. Gas emergency repair work is a high-risk job. There are many potential risk sources on site, such as gas leaks, fires, explosions, poisoning, etc. If these risks are not effectively identified and managed, they may lead to serious safety accidents. The following problems exist in the current gas emergency repair work:

[0003] First, research on gas emergency repairs has primarily focused on the concentration distribution patterns and hazardous area demarcation following a gas pipeline leak. Research on the gas emergency repair process has primarily focused on emergency material supply, emergency repair capabilities of repair teams, repair methods, and application scenarios. Research on the risk sources of gas emergency repairs has been limited, and it is not necessarily possible to fully and accurately identify risk sources.

[0004] Secondly, the various video surveillance systems are not interoperable, lacking a unified platform and interconnection under a cascade architecture;

[0005] Finally, the established video surveillance system is old and has a low level of intelligence. Summary of the Invention

[0006] In view of the above problems, an embodiment of the present invention provides an alarm method based on an AI video analysis algorithm to solve the problem in the prior art that risk sources may not be fully and accurately identified during gas emergency repairs, resulting in low gas emergency repair efficiency.

[0007] In a first aspect, an embodiment of the present invention provides an alarm method based on an artificial intelligence (AI) video analysis algorithm, which is applied to a smart terminal device. The method includes:

[0008] Obtaining media data sent by at least one interconnected device;

[0009] The media data is analyzed in real time through an AI video analysis algorithm; when an abnormal situation is detected, a master control alarm message is generated; the abnormal situation is used to indicate that the user has engaged in dangerous behavior or a safety hazard has occurred in the environment.

[0010] In one possible implementation, the AI video analysis algorithm includes a Faster R-CNN algorithm, and performing real-time analysis of the media data using the AI video analysis algorithm includes:

[0011] Build a Faster R-CNN model based on the region generation network RPN and Fast R-CNN;

[0012] The Faster R-CNN model is trained using sample data to obtain a Faster R-CNN training model.

[0013] The media data is analyzed in real time using the Faster R-CNN training model.

[0014] In one possible implementation, the AI video analysis algorithm includes a video analysis algorithm that combines target detection, target tracking, and behavior detection. The real-time analysis of the media data using the AI video analysis algorithm includes:

[0015] Detecting at least one target in the media data using the target detection algorithm and generating at least one target frame;

[0016] Tracking the at least one target using the target tracking algorithm to generate a target motion trajectory;

[0017] The action detection algorithm is used to perform action recognition on the target in the at least one target frame based on the target motion trajectory.

[0018] In a possible implementation, the master control alarm information includes at least one of smoking alarm information, leaving-office alarm information, smoke alarm information, flame alarm information, ground collapse alarm information, clothing alarm information, call alarm information and vehicle intrusion alarm information.

[0019] In a possible implementation, the interconnected device includes a wearable device and a video surveillance device, and the smart terminal device is communicatively connected to the wearable device and the video surveillance device respectively through wireless network communication technology.

[0020] In a possible implementation, the master control alarm information further includes face recognition alarm information, and the method further includes:

[0021] Through face recognition technology, the acquired face data to be identified is compared with the preset authorized face data. If the comparison fails, a face recognition alarm message is generated.

[0022] In a second aspect, an embodiment of the present invention provides a smart terminal device, the smart terminal device comprising:

[0023] The main control module is used to obtain media data sent by at least one connected device; perform real-time analysis of the media data using AI video analysis algorithms; and generate main control alarm information when an abnormality is detected; the abnormality is used to indicate dangerous user behavior or safety hazards in the environment;

[0024] A gateway module, configured to provide wireless network connection and local area network connection; and further configured to obtain first location information;

[0025] External interface module, used to connect to external devices to achieve communication transmission with external devices;

[0026] A display module is used to display status information through a user interface; it is also used to implement sound and light alarms according to the main control alarm information;

[0027] A power module is used to supply power to the smart terminal device.

[0028] In a possible implementation, the gateway module includes a router, an Ethernet unit, and a positioning unit;

[0029] The router is configured to provide a wireless network connection for at least one of the networked devices, and the networked device is configured to indicate a device connected to the router;

[0030] The Ethernet unit is used to realize the local area network connection between the external wired Ethernet device, the networked device and the main control module;

[0031] The positioning unit is used to obtain first position information.

[0032] In a third aspect, an embodiment of the present invention provides a wearable device, comprising:

[0033] An intelligent module, used to implement data processing in the wearable device;

[0034] The media data acquisition module includes an image acquisition unit and an audio acquisition unit, and is used to acquire image data through the image acquisition unit and audio data through the audio acquisition unit, and encode and compress the image data and the audio data before uploading them to the cloud platform;

[0035] A positioning module, configured to obtain second location information and upload the second location information to a cloud platform;

[0036] A wearing detection module, configured to detect whether a user is wearing the wearable device through a built-in infrared detection sensor; and generate a wearing detection alarm message when it is detected that the user is not wearing the wearable device;

[0037] The acceleration sensor module is used to determine the user's state based on the detected acceleration value and acceleration direction; when the user is in a fall state, a fall alarm message is generated;

[0038] The lighting module includes an illumination lamp and a laser lamp, and is used to provide supplementary lighting through the illumination lamp; the laser lamp emits laser light of a specific wavelength, measures the residual light intensity after the laser light is absorbed, and determines whether a gas leak has occurred based on the residual light intensity; and generates a gas leak alarm message when a gas leak occurs;

[0039] The alarm module includes a speaker and an indicator light, which is used to realize audible and visual alarms through the speaker and the indicator light according to the alarm information obtained;

[0040] A battery module is used to power the wearable device.

[0041] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the alarm method based on the AI video analysis algorithm as described in the first aspect or any possible implementation of the first aspect.

[0042] The technical solution provided by the embodiments of this invention utilizes AI video analysis algorithms to efficiently perform real-time analysis of media data. Based on the different abnormalities identified, different types of master control alarm information are promptly generated, accurately identifying the source of risk, improving the intelligence of alarms, and enhancing the efficiency of gas emergency repairs. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of an alarm method based on an AI video analysis algorithm provided in an embodiment of the present invention.

[0044] Figure 2 A schematic diagram of the structure of an intelligent terminal device provided by an embodiment of the present invention.

[0045] Figure 3 A schematic structural diagram of a wearable device provided by an embodiment of the present invention.

[0046] Figure 4 A schematic diagram of an alarm system based on an AI video analysis algorithm provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] To make the objectives, technical solutions, and advantages of the present invention more clear and understandable, the present invention is further described below with reference to the accompanying drawings and specific embodiments. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0048] Figure 1 A flowchart of an alarm method based on an AI video analysis algorithm provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method includes:

[0049] Step 101: Acquire media data sent by at least one interconnected device.

[0050] Each step in the embodiment of the present invention is executed by an intelligent terminal device.

[0051] In this step, the connected devices include wearable devices and video surveillance devices. The smart terminal device communicates with the wearable devices and video surveillance devices respectively via wireless network communication technology (i.e., Wi-Fi). The media data includes image data and audio data sent by the wearable devices, as well as video data sent by the video surveillance devices.

[0052] Step 102: Perform real-time analysis on media data using AI video analysis algorithms; when an abnormality is detected, generate a master control alarm message.

[0053] In this step, the main control alarm information includes at least one of smoking alarm information, leaving-office alarm information, smoke alarm information, flame alarm information, ground collapse alarm information, clothing alarm information, call alarm information and vehicle intrusion alarm information.

[0054] Specifically, when the smart terminal device detects that construction workers are smoking through AI video analysis algorithms, a smoking alarm message is generated; when it is detected that construction workers are not at their work posts, an off-duty alarm message is generated; when smoke is detected, a smoke alarm message is generated; when flames are detected, a flame alarm message is generated; when the ground is detected to have sunk and formed a sinkhole, a ground collapse alarm message is generated; when it is detected that construction workers are not wearing work clothes as required, a clothing alarm message is generated; when it is detected that construction workers are making or receiving phone calls, a call alarm message is generated; when it is detected that a construction vehicle may be subject to unauthorized entry or external interference, a vehicle intrusion alarm message is generated.

[0055] In this embodiment of the present invention, the master control alarm information also includes facial recognition alarm information. Using facial recognition technology, the acquired facial data to be identified is compared with pre-set facial data. If the comparison fails, a facial recognition alarm message is generated. This facial comparison verifies the identity of the individual, preventing intruders and improving construction site safety.

[0056] In the embodiments of the present invention, by generating different types of warning information, construction workers can be assisted in complying with safety regulations, reducing accident risks and improving construction safety at the construction site. Furthermore, it is convenient for management personnel to take appropriate safety measures based on different types of warning information, thereby reducing accident losses.

[0057] As an optional solution, the AI video analysis algorithm includes the Faster R-CNN algorithm. The AI video analysis algorithm performs real-time analysis on the media data, including: establishing a Faster R-CNN model based on the Region Proposal Network (RPN) and Fast R-CNN; training the Faster R-CNN model using sample data to obtain a Faster R-CNN training model; and performing real-time analysis on the media data using the Faster R-CNN training model.

[0058] Specifically, the sample data is determined by the intersection-over-union (IoU) ratio between the generated detection box and the target detection box. The sample data includes positive samples and negative samples. When the IoU ratio is less than a first threshold, the detection box is determined to be a negative sample. When the IoU ratio is greater than the first threshold and less than a second threshold, the detection box is determined to be an invalid sample and does not participate in the training of the Faster R-CNN model. When the IoU ratio is greater than the second threshold, the detection box is determined to be a positive sample. For example, the first threshold is 0.3 and the second threshold is 0.7.

[0059] In an embodiment of the present invention, the parameters of the Faster R-CNN training model are updated in the direction of decreasing the function value of the loss function until convergence conditions are reached. The loss function includes a classification loss function and a regression loss function.

[0060] As another alternative, AI video analysis algorithms include those that combine target detection, target tracking, and behavior detection. Using AI video analysis algorithms, media data is analyzed in real time, including: detecting at least one target in the media data using a target detection algorithm to generate at least one target frame; tracking at least one target using a target tracking algorithm to generate a target motion trajectory; and identifying the actions of the target within the at least one target frame based on the target motion trajectory using a behavior detection algorithm. For example, the target detection algorithm may be the Yolov5 algorithm, the target tracking algorithm may be the DeepSort algorithm, and the behavior detection algorithm may be the SlowFast algorithm.

[0061] Specifically, the SlowFast algorithm recognizes information in a video by simultaneously using two channels: a fast channel and a slow channel. The slow channel, with a lower sampling frequency, is used to capture spatial semantic information, effectively capturing relatively static information in the video. The fast channel, with a higher sampling frequency, is used to capture motion with fine temporal resolution, effectively capturing dynamic information in the video. By connecting the fast channel to the slow channel laterally, information fusion between the two channels is achieved, resulting in the final video information recognition result.

[0062] The technical solution provided by the embodiments of this invention utilizes AI video analysis algorithms to efficiently perform real-time analysis of media data. Based on the different abnormalities identified, different types of master control alarm information are promptly generated, accurately identifying the source of risk, improving the intelligence of alarms, and enhancing the efficiency of gas emergency repairs.

[0063] Figure 2 A schematic diagram of the structure of an intelligent terminal device provided by an embodiment of the present invention is shown in FIG. Figure 2 As shown, the intelligent terminal device includes a main control module 11, a gateway module 12, an external interface module 13, a display module 14, and a power module 15. The main control module 11 is communicatively connected to the gateway module 12, the external interface module 13, the display module 14, and the power module 15 respectively. The gateway module 12 is communicatively connected to the external interface module 13, and the external interface module 13 is electrically connected to the power module 15.

[0064] In an embodiment of the present invention, a main control module 11 is used to obtain media data sent by at least one interconnected device; perform real-time analysis of the media data using an AI video analysis algorithm; and generate a main control alarm message when an abnormality is detected. A gateway module 12 is used to provide wireless network connection and local area network connection; and is also used to obtain first location information. An external interface module 13 is used to connect to an external device to enable communication transmission with the external device. A display module 14 is used to display status information through a user interface; and is also used to implement audible and visual alarms based on the main control alarm message. A power supply module 15 is used to supply power to the smart terminal device. The first location information includes the location information of the smart terminal device and the location information of the wearable device; and the status information includes battery information, device number, network connection status, working status, and enterprise information.

[0065] In an embodiment of the present invention, the main control module 11 includes an AI edge computing unit 111. The AI edge computing unit 111 has a built-in high-performance graphics processor (GPU), which can realize up to 16 channels of network video access. Through edge computing technology, it can efficiently complete real-time analysis of media data and generate main control alarm information in a timely manner.

[0066] Optionally, the AI edge computing unit 111 is also used to perform face comparison between the acquired face data to be identified and the preset authorized face data through face recognition technology. If the comparison fails, a face recognition alarm message is generated.

[0067] In an embodiment of the present invention, the gateway module 12 includes a router 121, an Ethernet unit 122 and a positioning unit 123. The router 121 and the Ethernet unit 122 are connected to each other via a wide area network (WAN) interface. The router 121 is used to provide a wireless network connection for at least one networked device, and the networked device is used to indicate the device connected to the router. For example, the router is a 5G / 4G router that can support access to more than 50 devices and has a coverage range of up to 200m. The network standard defaults to a 5G network, and it also supports switching to a 4G network when the network environment is limited. Networked devices include wearable devices, video surveillance equipment, automated pipeline welding machines and other equipment at the construction site. The Ethernet unit 122 is used to realize a local area network connection between an external wired Ethernet device, a networked device and the main control module 11. The positioning unit 123 is used to obtain the first location information. The positioning unit 123 includes a global positioning system (GPS) subunit and a Beidou satellite positioning subunit.

[0068] In this embodiment of the present invention, the external interface module 13 includes a power interface 131, a Wi-Fi antenna interface 132, a mobile communication technology antenna interface 133, a network interface 134, a satellite positioning antenna interface 135, and a Universal Serial Bus (USB) interface 136. The display module 14 includes a speaker 141, an indicator light 142, and a display screen 143. The power module 15 includes a rechargeable battery 151 and a power management unit 152. The power interface 131 is electrically connected to the rechargeable battery 151, and an external battery charger is connected to the power interface 131 to charge the rechargeable battery 151. The rechargeable battery 151 is electrically connected to the power management unit 152 to provide power to the power management unit 152. The Wi-Fi antenna interface 132 and the mobile communication technology antenna interface 133 are respectively connected to the router via SMA interfaces, and the Wi-Fi antenna interface 132 and the mobile communication technology antenna interface 133 respectively provide external antennas for the router 121. The network interface 134 is connected to the Ethernet unit 122 via a local area network (LAN), and the satellite positioning antenna interface 135 is connected to the positioning unit 123 via an SMA interface. The satellite positioning antenna interface 135 provides an external antenna for the positioning unit 123. For example, the network interface is a gigabit network port, and the USB interface is a USB 3.0 interface.

[0069] In an embodiment of the present invention, the main control module 11 further includes a signal interface unit 112, and the AI edge computing unit 111 and the signal interface unit 112 are electrically connected via an internal bus. The signal interface unit 112 is connected to the Ethernet unit 122 via a LAN, and is connected to the positioning unit 123, the rechargeable battery 151, and the display screen 143 via an RS-485 serial port, and is connected to the USB interface 136 via USB 3.0 technology. In addition, the signal interface unit 112 is also used to control the lighting or extinguishing of the indicator light through the switch quantity, and to control the speaker 141 through the power amplifier to realize the sound and light alarm.

[0070] In the technical solution provided by the embodiment of the present invention, the smart terminal device serves as a local area network gateway, connected to the wearable device and video surveillance equipment respectively, and realizes real-time analysis of media data through a high-performance GPU. It can quickly and accurately identify the risk sources at the repair site, thereby improving the safety of the construction site and the efficiency of gas repair.

[0071] Figure 3 A structural diagram of a wearable device provided by an embodiment of the present invention is shown in FIG. Figure 3As shown, the wearable device includes: a smart module, a media data acquisition module, a positioning module, a wearing detection module, an accelerometer module, a lighting module, an alarm module, and a battery module. The hardware of the wearable device is designed based on the smart module. The smart module is a 5G smart module, including a super core with a main frequency of 2.7GHz, three large cores with a main frequency of 2.4GHz, and four small cores with a main frequency of 1.9GHz. The smart module has multiple interfaces, through which it is electrically connected to the media data acquisition module, positioning module, wearing detection module, accelerometer module, lighting module, alarm module, and battery module.

[0072] In an embodiment of the present invention, the intelligent module is used to realize data processing in the wearable device; the media data acquisition module includes an image acquisition unit and an audio acquisition unit, which are used to acquire image data through the image acquisition unit, acquire audio data through the audio acquisition unit, and encode and compress the image data and audio data and upload them to the cloud platform; the positioning module is used to obtain second position information and upload the second position information to the cloud platform. Among them, the image acquisition unit is a camera, and the audio acquisition unit is a microphone. In actual applications, the media data acquisition module can realize video call function and voice intercom function. In order to improve the quality of communication, a high-pixel camera can be used for image acquisition, and a microphone with dual-microphone noise reduction technology can be used for audio acquisition. The positioning module includes a GPS unit and a Beidou satellite positioning unit.

[0073] Specifically, after the positioning module uploads the second location information to the cloud platform in real time, the cloud platform sends the second location information to the computer device. The computer device generates an operation trajectory map based on the second location information, and the operation trajectory map can be used to intuitively understand the construction workers' operation trajectory.

[0074] Optionally, the computer device can also detect whether the second location information is within a preset danger zone; if it is detected that the second location information is within the preset danger zone, a danger zone alarm message is generated and sent to the wearable device. The wearable device implements an audible and visual alarm through the alarm module to remind construction personnel that they have entered the danger zone and to exit as soon as possible.

[0075] In this embodiment of the present invention, the wear detection module is used to detect whether the user is wearing a wearable device using a built-in infrared detection sensor. If the user is not wearing the wearable device, a wear detection alarm is generated. The acceleration sensor module is used to determine the user's state based on the detected acceleration value and acceleration direction; if the user is in a fall state, a fall alarm is generated.

[0076] Optionally, the wearable device can also send fall alarm information to computer equipment, so that back-end management personnel can take corresponding emergency measures in a timely manner according to the fall alarm information, thereby improving the safety of construction workers and reducing accident losses.

[0077] Specifically, the acceleration sensor module is configured to determine that the user is in a silent state when both the horizontal and vertical acceleration values are less than a preset acceleration threshold; determine that the user is in motion when the horizontal acceleration value is greater than the acceleration threshold and the acceleration value changes relatively smoothly; and determine that the user is falling when the vertical acceleration value suddenly increases. The horizontal and vertical directions are based on the ground as a reference, with the direction parallel to the ground defined as the horizontal direction and the direction perpendicular to the ground defined as the vertical direction.

[0078] In an embodiment of the present invention, the lighting module includes an illumination lamp and a laser lamp, which is used to provide fill light through the illumination lamp; the laser lamp emits laser light of a specific wavelength, measures the residual light intensity after the laser light is absorbed, and determines whether a gas leak occurs based on the residual light intensity; when a gas leak occurs, a gas leak alarm message is generated.

[0079] In an embodiment of the present invention, the alarm module includes a speaker and an indicator light, and is configured to provide an audible and visual alarm based on the acquired alarm information. The battery module is configured to power the wearable device. The alarm information acquired by the alarm module can be alarm information generated internally by the wearable device or master control alarm information sent by a smart terminal device. Specifically, upon acquiring at least one of a wear detection alarm, a fall alarm, a gas leak alarm, or a master control alarm, the alarm module provides an audible and visual alarm through the speaker and indicator light.

[0080] In the technical solution provided by the embodiment of the present invention, wearable devices can cooperate with computer equipment to remotely track the dynamics of on-site workers, conduct real-time monitoring and work guidance, and realize real-time linkage between on-site operations and background management, solving the problem of real-time monitoring of on-site workers and improving management efficiency.

[0081] In an embodiment of the present invention, the wearable device is an intelligent safety helmet that has the functions of wearing detection, personnel positioning, fall detection, video / voice call, gas detection, and sound and light alarm, which improves the intelligence level of the safety helmet and improves the safety of the construction site.

[0082] Figure 4 A schematic diagram of an alarm system based on an AI video analysis algorithm provided by an embodiment of the present invention is shown in FIG. Figure 4 As shown, the system includes: application layer, platform layer and algorithm layer.

[0083] The application layer includes the cloud platform, computer devices, and mobile terminals. The cloud platform is connected to the computer devices and mobile terminals respectively. The computer devices use installed background management software to store and analyze data and issue control instructions. The mobile terminals use installed applications to monitor the status of each device in the platform layer and control the mode.

[0084] In an embodiment of the present invention, communication transmission between the application layer and the platform layer is achieved through mobile communication technology based on the Message Queuing Telemetry Transport (MQTT) protocol. The MQTT protocol is an International Organization for Standardization (ISO) standard protocol that works on the Transmission Control Protocol / Internet Protocol (TCP / IP) protocol. Its low overhead, low bandwidth occupancy, and instant messaging features allow it to provide real-time and reliable message services for connected remote devices in low-bandwidth, high-latency or unreliable network environments with very little code and very low power consumption. At the same time, its versatility makes it very easy for terminal systems based on the MQTT protocol to expand device types and business types, and it can also be directly connected to open cloud platforms that are also based on the protocol.

[0085] In an embodiment of the present invention, the platform layer includes a smart terminal device, at least one wearable device, and at least one video surveillance device. Each device in the platform layer can communicate with the cloud platform via mobile communication technology (such as fourth-generation mobile communication technology, i.e., 4G, or fifth-generation mobile communication technology, i.e., 5G). The smart terminal device, at least one wearable device, and at least one video surveillance device are connected to each other via Wi-Fi. The wearable device includes a smart helmet.

[0086] In embodiments of the present invention, smart terminal devices can obtain the status of access devices and implement local security management of access devices. Access devices refer to devices connected to smart terminal devices. Optionally, the smart terminal devices transmit access device data to a cloud platform, which in turn transmits the access device data to a computer device and a mobile terminal, allowing administrators to control and manage the access devices via the computer device or mobile terminal.

[0087] Optionally, the cloud platform sends a mode control instruction to the smart terminal device, and the smart terminal device switches the working mode in response to the mode control instruction.

[0088] In the embodiment of the present invention, the algorithm layer includes a smoking recognition algorithm, an off-duty recognition algorithm, a smoke recognition algorithm, a flame recognition algorithm, a ground collapse recognition algorithm, a clothing recognition algorithm, a call recognition algorithm and a vehicle intrusion recognition algorithm.

[0089] Optionally, when detecting ground collapse, a ground collapse recognition algorithm can be used in combination with ground collapse sensors installed at the construction site to detect ground collapse, so as to improve the accuracy of ground collapse alarms and reduce missed or false alarms.

[0090] In the technical solution provided by the embodiment of the present invention, the smart terminal device serves as a local area network gateway, connected to the wearable device and video surveillance equipment respectively, and realizes real-time analysis of media data through a high-performance GPU. It can quickly and accurately identify the risk sources at the repair site, thereby improving the safety of the construction site and the efficiency of gas repair.

[0091] In the embodiment of the present invention, the computer equipment can quickly respond to the main control alarm information and take corresponding emergency measures in time, thereby improving the safety of the construction site and reducing accident losses.

[0092] In an embodiment of the present invention, an alarm system based on an AI video analysis algorithm is established, which realizes the interconnection between the application layer and the platform layer, between the various devices in the application layer, and between the various devices in the platform layer, solves the problem of lack of interconnection under the cascade architecture in related technologies, and can realize unified management of wearable devices, smart terminal devices and video surveillance equipment, thereby improving the intelligence level of the construction site and improving management efficiency.

[0093] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program. When the program is running, the device where the computer-readable storage medium is located is controlled to execute the steps of the embodiment of the above-mentioned alarm method based on the AI video analysis algorithm. For a specific description, please refer to the embodiment of the above-mentioned alarm method based on the AI video analysis algorithm.

[0094] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An alarm method based on artificial intelligence (AI) video analysis algorithm, characterized in that: Applied to a smart terminal device, the method includes: Obtaining media data sent by at least one interconnected device; The media data is analyzed in real time through an AI video analysis algorithm; when an abnormal situation is detected, a master control alarm message is generated; the abnormal situation is used to indicate that the user has engaged in dangerous behavior or a safety hazard has occurred in the environment.

2. The method according to claim 1, characterized in that The AI video analysis algorithm includes a Faster R-CNN algorithm. The AI video analysis algorithm is used to perform real-time analysis on the media data, including: Build a Faster R-CNN model based on the region generation network RPN and Fast R-CNN; The Faster R-CNN model is trained using sample data to obtain a Faster R-CNN training model. The media data is analyzed in real time using the Faster R-CNN training model.

3. The method according to claim 1, characterized in that The AI video analysis algorithm includes a video analysis algorithm that combines target detection, target tracking, and behavior detection. The AI video analysis algorithm is used to perform real-time analysis of the media data, including: Detecting at least one target in the media data using the target detection algorithm and generating at least one target frame; Tracking the at least one target using the target tracking algorithm to generate a target motion trajectory; The action detection algorithm is used to perform action recognition on the target in the at least one target frame based on the target motion trajectory.

4. The method according to claim 1, wherein The main control alarm information includes at least one of smoking alarm information, leaving-office alarm information, smoke alarm information, flame alarm information, ground collapse alarm information, clothing alarm information, call alarm information and vehicle intrusion alarm information.

5. The method according to claim 1, wherein The interconnected devices include a wearable device and a video surveillance device, and the smart terminal device is respectively connected to the wearable device and the video surveillance device through wireless network communication technology.

6. The method according to claim 4, characterized in that The master control alarm information also includes face recognition alarm information, and the method further includes: Through face recognition technology, the acquired face data to be identified is compared with the preset authorized face data. If the comparison fails, a face recognition alarm message is generated.

7. An intelligent terminal device, characterized in that: The intelligent terminal device includes: The main control module is used to obtain media data sent by at least one connected device; perform real-time analysis of the media data using AI video analysis algorithms; and generate main control alarm information when an abnormality is detected; the abnormality is used to indicate dangerous user behavior or safety hazards in the environment; A gateway module, configured to provide wireless network connection and local area network connection; and further configured to obtain first location information; External interface module, used to connect to external devices to achieve communication transmission with external devices; A display module is used to display status information through a user interface; it is also used to implement sound and light alarms according to the main control alarm information; A power module is used to supply power to the smart terminal device.

8. The intelligent terminal device according to claim 7, characterized in that: The gateway module includes a router, an Ethernet unit and a positioning unit; The router is configured to provide a wireless network connection for at least one of the networked devices, and the networked device is configured to indicate a device connected to the router; The Ethernet unit is used to realize the local area network connection between the external wired Ethernet device, the networked device and the main control module; The positioning unit is used to obtain first position information.

9. A wearable device, characterized in that: The wearable device includes: An intelligent module, used to implement data processing in the wearable device; The media data acquisition module includes an image acquisition unit and an audio acquisition unit, and is used to acquire image data through the image acquisition unit and audio data through the audio acquisition unit, and encode and compress the image data and the audio data before uploading them to the cloud platform; A positioning module, configured to obtain second location information and upload the second location information to a cloud platform; A wearing detection module, configured to detect whether a user is wearing the wearable device through a built-in infrared detection sensor; and generate a wearing detection alarm message when it is detected that the user is not wearing the wearable device; The acceleration sensor module is used to determine the user's state based on the detected acceleration value and acceleration direction; when the user is in a fall state, a fall alarm message is generated; The lighting module includes an illumination lamp and a laser lamp, and is used to provide supplementary lighting through the illumination lamp; the laser lamp emits laser light of a specific wavelength, measures the residual light intensity after the laser light is absorbed, and determines whether a gas leak has occurred based on the residual light intensity; and generates a gas leak alarm message when a gas leak occurs; The alarm module includes a speaker and an indicator light, which is used to realize audible and visual alarms through the speaker and the indicator light according to the alarm information obtained; A battery module is used to power the wearable device.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein, when the program is running, the device where the computer-readable storage medium is located is controlled to execute the alarm method based on the AI video analysis algorithm as described in any one of claims 1 to 6.

Images