Real-time data transmission and cloud storage management system for motion camera

Abstract

The invention relates to the technical field of data communication and processing, and discloses a motion camera real-time data transmission and cloud storage management system, which comprises a semantic coding module used for processing real-time sensor data; the first communication module is used for sending the semantic event to a cloud service platform; the data extraction decision module is used for receiving the semantic event and generating a data extraction trigger signal according to a preset condition; the reverse scheduling module is used for responding to the data extraction trigger signal to generate a data extraction instruction containing information associated with the timestamp; and the second communication module is used for acquiring a corresponding video data block from a local storage medium after the motion camera receives the data extraction instruction. According to the method, the semantic event is reported through the low-power-consumption channel, and the reverse scheduling instruction of the cloud is received, so that high-value video clips are accurately extracted as required under the conditions of low power consumption and low bandwidth consumption.

Description

Technical Field

[0001] This invention relates to the field of data communication and processing technology, specifically to a real-time data transmission and cloud storage management system for action cameras. Background Technology

[0002] Action cameras are widely used for recording various sports activities and outdoor adventures due to their portability and ruggedness. In these applications, the devices typically record continuously and generate a large number of raw video files, often in gigabytes (GB), containing long periods of continuous footage.

[0003] Current technologies for acquiring and managing video content have significant shortcomings. A common approach is to transfer all video files stored in the camera to a personal computer or other processing device at once, either via wired connection or by manually inserting and removing memory cards, after the shooting activity. This method suffers from high latency, preventing users from instantly reviewing and sharing highlights during or immediately after the activity, and requiring them to spend considerable time manually searching for and editing valuable segments from lengthy videos.

[0004] To improve the immediacy of content, some existing solutions employ real-time streaming or full file upload functionality based on wireless networks (such as Wi-Fi). Users can view the footage in real time via mobile devices or wirelessly transfer files to the cloud after recording. However, this method requires the action camera to continuously keep its high-power Wi-Fi module powered on to maintain high-bandwidth data transmission, which shortens the device's effective battery life outdoors—a major limitation for portable cameras with limited battery capacity.

[0005] Furthermore, this method is highly dependent on the network environment. In many sports scenarios, such as remote mountainous areas or water sports environments, unstable wireless network signals or insufficient bandwidth can lead to interruptions in real-time streaming or severe degradation in image quality. The entire file upload process can also become extremely slow or frequently fail. Even when network conditions permit, uploading all raw video data indiscriminately to the cloud fails to solve the technical problem of quickly locating key moments from massive amounts of data. It merely shifts the arduous filtering work from local to cloud environments, without fundamentally optimizing the efficiency of discovering and acquiring high-value content. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a real-time data transmission and cloud storage management system for action cameras. This system solves the problem that existing action camera technologies, when realizing the instant discovery and on-demand acquisition of high-value video clips, suffer from continuous reliance on high-power communication modules and strong dependence on high-bandwidth networks, resulting in shortened device battery life and low data transmission efficiency.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a real-time data transmission and cloud storage management system for action cameras, comprising: The semantic encoding module is used to process real-time sensor data to generate semantic events containing timestamps; The first communication module is used to send the semantic event to the cloud service platform and receive instructions from the cloud service platform; The data extraction decision module is used to receive the semantic event and generate a data extraction trigger signal according to preset conditions; The reverse scheduling module is used to respond to the data extraction trigger signal to generate a data extraction instruction containing information associated with the timestamp, and to send the data extraction instruction to the first communication module; The second communication module is used to retrieve the corresponding video data block from the local storage medium after the action camera receives the data extraction instruction, and upload the video data block to the cloud service platform through a high-bandwidth channel.

[0008] In one specific embodiment, the semantic encoding module includes: a data acquisition unit for acquiring video streams from the motion camera, inertial measurement unit data streams, and audio streams; a multimodal fusion analysis unit for taking the video streams, inertial measurement unit data streams, and audio streams as inputs and processing them in real time through a preset edge AI coprocessor; and an event generation unit for generating the semantic event when the multimodal fusion analysis unit identifies a preset action event, visual event, or audio event, wherein the semantic event contains a precise timestamp of the event occurrence point.

[0009] Preferably, the first communication module includes: a channel selection unit for selecting a low-power, low-bandwidth heartbeat channel as a data transmission carrier; an event reporting unit for sending the semantic metadata stream encapsulated in the semantic event to the cloud service platform through the heartbeat channel; and an instruction receiving unit for receiving the data extraction instruction from the cloud service platform through the heartbeat channel.

[0010] Furthermore, the instruction receiving unit is specifically used to receive the data extraction instruction through the heartbeat channel. After the data extraction instruction is generated, it is directly transmitted through the heartbeat channel to ensure that the motion camera can still receive the instruction when the second communication module is not in an active state.

[0011] In one specific embodiment, the data extraction decision module includes: a semantic parsing unit, used to receive the semantic event and parse its attribute information into a queryable data structure; a rule matching unit, used to determine whether the semantic event satisfies a preset semantic subscription rule, wherein the semantic subscription rule is composed of one or more matching predicate logical conjunctions related to the event attributes; and a trigger determination unit, used to perform calculations on the determination result of the rule matching unit and the viewing request from the user terminal to generate the data extraction trigger signal.

[0012] Furthermore, the triggering determination unit is specifically used to perform a logical OR operation between the determination result of the rule matching unit and the determination result of the viewing request from the user terminal, and generate the data extraction trigger signal when the semantic subscription rule is matched or the viewing request is received.

[0013] Preferably, the reverse scheduling module includes: an instruction parameter calculation unit, used to calculate the start timestamp and duration of the video data block required by the data extraction instruction based on the timestamp contained in the semantic event, combined with preset advance buffer and lag buffer amounts; an instruction generation unit, used to encapsulate the start timestamp, the duration, the unique identifier of the target device, and the data quality specification into the data extraction instruction; and an instruction issuing unit, used to send the data extraction instruction to the motion camera end through the first communication module.

[0014] Furthermore, the instruction parameter calculation unit is specifically used to calculate the start timestamp by subtracting a preset advance buffer amount from the timestamp of the semantic event, and to calculate the duration by adding the preset advance buffer amount and the preset lag buffer amount to the duration of the semantic event.

[0015] In one specific embodiment, the second communication module includes: a high-bandwidth channel wake-up unit, used to wake up a high-bandwidth channel that is in a sleep or low-power state by default after receiving the data extraction instruction; a data positioning and reading unit, used to accurately locate and read the video data block from the local storage medium according to the start timestamp and duration contained in the data extraction instruction; and a high-speed upload unit, used to upload the video data block to the cloud service platform through the high-bandwidth channel, and to return the high-bandwidth channel to a sleep or low-power state after the upload is completed.

[0016] A second aspect of this invention provides a method for real-time data transmission and cloud storage management of an action camera. This method, applied to the aforementioned real-time data transmission and cloud storage management system for action cameras, includes the following steps: S1. The semantic encoding module processes real-time sensor data to generate semantic events containing timestamps. S2. The semantic event is sent to the cloud service platform through the first communication module, and the data extraction decision module generates a data extraction trigger signal based on the received semantic event and preset conditions. S3. The reverse scheduling module responds to the data extraction trigger signal to generate a data extraction instruction containing information associated with the timestamp, and sends the data extraction instruction to the motion camera through the first communication module. S4. After the action camera receives the data extraction instruction, the second communication module obtains the corresponding video data block from the local storage medium and uploads the video data block to the cloud service platform through a high-bandwidth channel.

[0017] Preferably, in step S3, the step of generating the data extraction instruction includes: encapsulating the unique identifier of the target device, the start timestamp calculated based on the timestamp, the duration of the video data block to be extracted, and the data quality specifications into the data extraction instruction.

[0018] This invention provides a real-time data transmission and cloud storage management system for action cameras. It offers the following advantages: 1. This invention achieves real-time semantic indexing of video content by setting a semantic encoding module on the action camera to generate lightweight semantic events in real time, and then using a first communication module to send these semantic events to the cloud in advance. Users or cloud services can instantly know the occurrence time and type of high-value segments without waiting for the large original video file to be transmitted, thereby reducing the delay from shooting to discovering high-value content.

[0019] 2. This invention achieves on-demand use of high-power, high-bandwidth channels by setting up independent first and second communication modules for transmitting semantic events and video data blocks respectively, and combining them with a cloud-based reverse scheduling mechanism. The second communication module is only activated for brief transmission when a specific video data block is needed; otherwise, the connection is maintained solely by the low-power first communication module. This significantly reduces the overall power consumption of the action camera during real-time interaction and effectively extends its outdoor battery life.

[0020] 3. This invention, by setting up a data extraction decision module in the cloud, allows data extraction to be triggered based on user requests or preset semantic subscription rules, and precisely sends the specific extraction range to the camera end through instructions generated by the reverse scheduling module. This collaborative working mode, led by the cloud and precisely executed by the camera end, avoids invalid or low-quality transmission of entire video streams in motion scenarios with limited network bandwidth, ensuring that users ultimately obtain the original, lossless, high-definition video clips they are interested in. Attached Figure Description

[0021] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figure 1 , Figure 1 This is an architecture diagram of a real-time data transmission and cloud storage management system for an action camera according to an embodiment of the present invention. This embodiment provides a real-time data transmission and cloud storage management system for an action camera, which includes an action camera configured on the user side and a remotely deployed cloud service platform. The action camera and the cloud service platform communicate via a wireless network.

[0024] An action camera internally includes a processor, memory, communication hardware, local storage media, and multiple sensors, such as an image sensor and an inertial measurement unit. The action camera also includes a semantic encoding module, a first communication module, and a second communication module to implement its functions.

[0025] The cloud service platform can be one or a group of physical servers or virtual server clusters, on which the following modules are deployed: data extraction and decision-making module and reverse scheduling module.

[0026] In this embodiment, the semantic encoding module processes real-time sensor data acquired by the action camera to generate semantic events containing timestamps. A semantic event is a structured description of a specific segment of the video content.

[0027] The first communication module establishes and maintains a low-power, low-bandwidth channel, referred to in this embodiment as the heartbeat channel. This heartbeat channel is used to report semantic events generated by the semantic encoding module to the cloud service platform and to receive data extraction instructions issued by the cloud service platform. Due to its low-power characteristics, this heartbeat channel can remain continuously online after the action camera is powered on.

[0028] The second communication module is used to establish and maintain a high-bandwidth channel. This high-bandwidth channel is in a dormant or inactive state by default. The second communication module is only awakened and activated when the action camera receives a valid data extraction command through the first communication module, so as to upload the video data block specified by the command, stored in the local storage medium, to the cloud service platform at high speed.

[0029] On the cloud service platform side, the data extraction decision module is used to receive and parse the semantic events reported by the first communication module, and determine whether it is necessary to obtain the original video data block corresponding to the semantic event based on preset conditions, thereby generating a data extraction trigger signal.

[0030] The reverse scheduling module is connected to the data extraction decision module. When a data extraction trigger signal is received, the reverse scheduling module generates a data extraction instruction containing precise extraction parameters and sends the instruction to the motion camera via the heartbeat channel, thereby completing the remote data extraction scheduling for the motion camera.

[0031] Reference Figure 1 In this embodiment, the semantic encoding module is configured on the hardware circuit inside the action camera, and the specific functions of the semantic encoding module are implemented through the collaborative efforts of multiple internal units.

[0032] The semantic coding module includes a data acquisition unit, a multimodal fusion analysis unit, and an event generation unit.

[0033] The data acquisition unit connects to multiple internal sensors of the action camera for parallel, real-time sensor data acquisition. Specifically, it acquires a video stream from the image sensor, an inertial measurement unit (IMU) data stream containing triaxial acceleration and angular velocity, and an audio stream from the microphone array. These data streams are tagged with synchronized timestamps for subsequent processing.

[0034] The input of the multimodal fusion analysis unit is connected to the output of the data acquisition unit to receive video streams, inertial measurement unit data streams, and audio streams. This multimodal fusion analysis unit runs a pre-defined analysis model on a dedicated edge AI coprocessor within the action camera. This analysis model is trained to recognize specific temporal patterns in the data streams, corresponding to pre-defined action events (e.g., jumping, turning), visual events (e.g., specific gestures, facial smiles), or audio events (e.g., specific command words). The multimodal fusion analysis unit processes the input multimodal data in real time, and when it identifies data segments that match the pre-defined patterns, it outputs recognition results including event categories and confidence scores.

[0035] The input of the event generation unit is connected to the output of the multimodal fusion analysis unit. Upon receiving the recognition result, the event generation unit constructs a structured semantic event. This semantic event is a quantitative description of the recognized event. In this embodiment, the semantic event... Defined as a tuple, its data structure is as follows: ; In the formula, A precise timestamp representing the point in time when the event occurred. This timestamp is determined by the multimodal fusion analysis unit and aligned with the frame timestamps of the video stream; The estimated duration of the event is represented by a numerical value, indicating the time span from the start to the end of the event; The geographic coordinates representing the time the event occurred are derived from the GPS module built into the action camera. A discrete type identifier representing an event, which is a predefined enumeration value used to distinguish different event categories; The quantitative intensity or confidence score representing the event is a real number output by the multimodal fusion analysis unit, indicating the reliability of the identification result or the severity of the event; This represents an optional additional metadata field used to store other information related to the event, such as the IDs of the people involved in the event or environmental parameters.

[0036] After generating a semantic event, the event generation unit outputs it to the first communication module for subsequent processing.

[0037] Reference Figure 1 In this embodiment, the first communication module is configured as a functional module combining hardware and firmware inside the action camera, responsible for managing low-power bidirectional communication with the cloud service platform.

[0038] The first communication module includes a channel selection unit, an event reporting unit, and an instruction receiving unit.

[0039] The channel selection unit scans and selects a low-power, low-bandwidth wireless communication protocol after the action camera is activated, and establishes a connection with the cloud service platform based on this protocol. This connection is referred to as a heartbeat channel in this embodiment. Communication protocols that can be used to establish a heartbeat channel include, but are not limited to, Bluetooth Low Energy or long-range radio. After the connection is established, the channel selection unit is responsible for maintaining the continuous availability of the heartbeat channel.

[0040] The input of the event reporting unit is connected to the output of the semantic encoding module. This unit receives a series of semantic events generated by the semantic encoding module and encapsulates them into a semantic metadata stream for transmission. Specifically, the event reporting unit organizes multiple semantic events arriving in chronological order into a sequence. Semantic metadata stream It can be represented as: ; In the formula, Represents the encapsulated semantic metadata stream; Represents the first in the semantic metadata stream semantic events, among which It is from 1 to Integer index; Represents a semantic metadata stream encapsulated in one go. The total number of semantic events contained therein.

[0041] Before transmission, the event reporting unit encodes the serialized data structure into a byte stream using a format such as JSON (JavaScript Object Notation) or Protobuf (Protocol Buffers). This byte stream is then sent to the cloud service platform via a heartbeat channel established by the channel selection unit.

[0042] The command receiving unit continuously monitors downlink data from the cloud service platform on the heartbeat channel. When a data packet is received, the command receiving unit parses it to determine if it is a valid data extraction command. Because the command receiving unit shares the same always-on heartbeat channel with the event reporting unit, the action camera can still receive and process scheduling commands issued from the cloud even if the second communication module is inactive. After successfully parsing and verifying a data extraction command, the command receiving unit transmits the command content to the action camera's central processing unit to trigger the subsequent data extraction and upload process.

[0043] Reference Figure 1 In this embodiment, the data extraction and decision-making module is deployed on a cloud service platform to analyze and make decisions on reported semantic events.

[0044] The data extraction decision module includes a semantic parsing unit, a rule matching unit, and a trigger determination unit.

[0045] The input of the semantic parsing unit is connected to the reporting path of the first communication module to receive a byte stream encapsulated as semantic metadata. The semantic parsing unit first deserializes the received byte stream according to a preset encoding format (e.g., JSON or Protobuf), restoring it to a data structure containing one or more semantic events. Subsequently, the semantic parsing unit stores each parsed semantic event and its contained attribute information in a database or memory cache that can be queried in real time, and provides a query interface for the user terminal to generate a semantic timeline on the user interface.

[0046] The input of the rule matching unit is connected to the output of the semantic parsing unit. This rule matching unit determines whether each newly arriving semantic event satisfies the semantic subscription rules stored in the cloud service platform. Semantic subscription rules. With authorized users An association is formed by logical conjunction of one or more matching predicates. An event. Determined to meet the rules The condition is that the event makes all matching predicates included in the rule true. This matching relation can be represented as: ; In the formula, It is a Boolean function whose value of true indicates an event. Satisfy users Subscription rules ; Representing the A matching predicate, which is an action applied to an event. Boolean functions for attributes; This represents the logical AND operation; `<index>` is an integer index variable. It is used to iterate through the subscription rules. Each matching predicate has a value ranging from 1 to ; This represents a specific subscription rule. The total number of matching predicates; It is a Boolean constant representing logical "true".

[0047] In this embodiment, matching predicates This may include, but is not limited to: Type predicate: determines the type of an event. Does it belong to the set of user subscription types?

[0048] Intensity predicate: Used to determine the intensity of an event. Is it greater than or equal to the intensity threshold set by the user?

[0049] Geographic location predicate: determines the location of an event. Is it within a user-defined geofence area?

[0050] The input of the trigger determination unit is connected to the output of the rule matching unit. Simultaneously, this trigger determination unit also receives viewing requests from the user terminal. The trigger determination unit generates the final data extraction trigger signal by performing a logical OR operation on two trigger conditions. Specifically, for a newly arrived event... Its final trigger signal The generation logic is as follows: ; In the formula, It is in response to the event. The final data extraction trigger signal is a Boolean value; This represents the collection of all authorized users associated with this action camera; Represents the logical OR operation; It is a Boolean function whose value of true indicates an event. Satisfy users Subscription rules ; It is a boolean function, when the user Events displayed on the screen via its user terminal Its value is true when a request to view high-definition clips is initiated; For users; Operators representing set ownership relationships.

[0051] Reference Figure 1 In this embodiment, the reverse scheduling module is deployed on the cloud service platform. Its input end is connected to the output end of the data extraction decision module. After receiving the data extraction trigger signal, it generates and issues specific data extraction instructions.

[0052] The reverse scheduling module includes an instruction parameter calculation unit, an instruction generation unit, and an instruction issuance unit.

[0053] The instruction parameter calculation unit is used to receive signals sent by the data extraction and decision module, which contain the semantic events that triggered this scheduling. The instruction parameter calculation unit calculates the core parameters required to generate the data extraction instruction based on the attributes of the semantic event. Specifically, the instruction parameter calculation unit calculates the start timestamp of the video data block to be extracted. Total duration The calculation method is as follows: ; ; In the formula, This represents the start timestamp of the video data block to be extracted in the local storage medium of the action camera; This represents the total duration of the video data block to be extracted; It is the precise timestamp of the event occurrence point recorded in the semantic event; It is the estimated duration of the event recorded in the semantic event; It is a pre-set buffer amount in the cloud service platform, which is a time value used to ensure that the extracted data contains the context before the event occurred; It is a preset lag buffer in the cloud service platform, which is a time value used to ensure that the extracted data includes the content after the event has ended.

[0054] After the calculation is completed, the instruction parameter calculation unit will calculate the result. and Output to the instruction generation unit.

[0055] The instruction generation unit is used to encapsulate the calculated parameters and other necessary scheduling information into a structured data extraction instruction. In this embodiment, the instruction is defined as a tuple as follows: ; In the formula, A unique device identifier representing the target motion camera, used for addressing in multi-device scenarios; and The start timestamp and duration are calculated by the instruction parameter calculation unit; The data quality specification represents the requested data, which may include parameters such as resolution, frame rate, and encoding format. It can be configured by system default or specified in the user's semantic subscription rules.

[0056] The input terminal of the instruction issuing unit is connected to the output terminal of the instruction generating unit. This instruction issuing unit receives the encapsulated data extraction instruction. The data is serialized into a byte stream and sent to the target motion camera via the low-power heartbeat channel established by the first communication module through the communication interface of the cloud service platform. Sending commands through this path ensures that the motion camera can receive scheduling commands even when its second communication module is inactive.

[0057] Reference Figure 1 In this embodiment, the second communication module is a hardware and firmware combined functional module configured inside the action camera, used to perform on-demand high-speed data upload tasks.

[0058] The second communication module includes a high-bandwidth channel wake-up unit, a data positioning and reading unit, and a high-speed upload unit.

[0059] The control input terminal of the high-bandwidth channel wake-up unit is connected to the central processing unit of the motion camera. When the motion camera receives and parses the data extraction command through the first communication module... Subsequently, the high-bandwidth channel wake-up unit receives a wake-up signal. In response to this signal, the high-bandwidth channel wake-up unit switches the high-bandwidth communication hardware (such as a Wi-Fi chipset) that is in a sleep or low-power state by default to an active state, and establishes a connection with the wireless access point according to the preset network configuration, thereby establishing a high-bandwidth channel.

[0060] The data positioning and reading unit is activated after the high-bandwidth channel is successfully established. This data positioning and reading unit retrieves data from the data extraction command. Get the start timestamp and duration It accesses video files stored on the action camera's local storage medium. By parsing the index information of the video container file (e.g., an MP4 file), this data is located and read by the timestamp. This maps to a specific byte offset within the file. Then, starting from that byte offset, a segment of data is read sequentially, with the time corresponding to the duration of the segment. The read data constitutes a video data block. This video data block can be defined as: ; In the formula, This represents the video data block that has been located and read. Represents the first in the video stream Frame image; Representing the The display timestamp of the frame image; and These are data extraction commands. The start time and duration specified in the settings.

[0061] The input of the high-speed upload unit is connected to the output of the data positioning and reading unit. This high-speed upload unit receives video data blocks. The data is then uploaded to the designated receiving address of the cloud service platform via the high-bandwidth channel established by the high-bandwidth channel wake-up unit. After the data transmission is complete and a successful transmission confirmation is received from the cloud service platform, the high-speed upload unit sends a sleep signal to the high-bandwidth channel wake-up unit. In response to this sleep signal, the high-bandwidth channel wake-up unit disconnects from the wireless access point and returns the high-bandwidth communication hardware to a sleep or low-power state, thus completing the full lifecycle of an on-demand upload and conserving energy.

[0062] Reference Figure 2 , Figure 2 This is a flowchart illustrating a real-time data transmission and cloud storage management method for an action camera according to an embodiment of the present invention. The method of this embodiment is implemented based on the system architecture described in the foregoing embodiments, and its specific process includes the following steps: Step S100: The user interacts with the cloud service platform through a user terminal (e.g., a smartphone application). At this stage, the user can configure one or more semantic subscription rules. These rules are then associated with the user's account and stored on the cloud service platform. A semantic subscription rule defines the type of event, intensity threshold, or geographical area that the user is interested in. Simultaneously, the action camera powers on, the first communication module initializes, and establishes and maintains a low-power heartbeat channel with the cloud service platform. The action camera begins recording video, and its internal semantic encoding module performs real-time analysis of the acquired multimodal sensor data (video stream, inertial measurement unit data stream, and audio stream) in parallel. When an event matching a preset pattern is identified, a structured semantic event is generated, containing information such as a precise timestamp, event type, and duration. Subsequently, the semantic event is encapsulated into a semantic metadata stream and sent to the cloud service platform through the heartbeat channel established by the first communication module.

[0063] Step S200: After receiving the semantic metadata stream, the data extraction and decision-making module of the cloud service platform parses it into semantic events. The data extraction decision module performs two parallel judgments: first, the rule matching unit judges the event. Does it meet the semantic subscription rules configured by any user in step S100? Secondly, the trigger determination unit monitors whether a user has accessed the event via a terminal. (Already displayed on the terminal interface) A manual viewing request has been initiated. When at least one of the above two conditions is met, the trigger determination unit generates a data extraction trigger signal.

[0064] Step S300: Upon receiving the data extraction trigger signal, the reverse scheduling module of the cloud service platform is activated. The instruction parameter calculation unit of this reverse scheduling module calculates the instruction parameter based on the trigger event. The timestamp and duration are used, along with preset advance and lag buffer amounts, to calculate the precise start timestamp of the video data block to be extracted. Total duration Subsequently, the instruction generation unit encapsulates these parameters, along with the target device's unique identifier and data quality specifications, into a data extraction instruction. The command is sent to the target motion camera via the heartbeat channel.

[0065] Step S400: The first communication module of the action camera receives the data extraction command. The data is then analyzed. After successful verification, the central processing unit controls the second communication module. The high-bandwidth channel wake-up unit is activated, waking up the high-bandwidth communication hardware (such as Wi-Fi) from sleep mode and establishing a connection. The data positioning and reading unit then executes the commands according to the instructions. and Accurately locate and read the corresponding video data blocks from the local storage medium. .

[0066] Step S400: Video data block The data block is uploaded to the cloud service platform via the activated high-bandwidth channel by the high-speed upload unit. The cloud service platform receives the data block and compares it with previously received semantic events. The video data block is then associated and stored. After the upload is complete, the action camera's second communication module returns to sleep mode. For users who initiate a manual request, the video data block is pushed to their user terminal for playback; for data blocks that are automatically uploaded due to meeting subscription rules, the cloud stores them and marks them as high-value segments for users to review later.

Claims

1. A real-time data transmission and cloud storage management system for action cameras, characterized in that, include: The semantic encoding module is used to process real-time sensor data to generate semantic events containing timestamps; The first communication module is used to send the semantic event to the cloud service platform and receive instructions from the cloud service platform; The data extraction decision module is used to receive the semantic event and generate a data extraction trigger signal according to preset conditions; The reverse scheduling module is used to respond to the data extraction trigger signal to generate a data extraction instruction containing information associated with the timestamp, and to send the data extraction instruction to the first communication module; The second communication module is used to retrieve the corresponding video data block from the local storage medium after the action camera receives the data extraction instruction, and upload the video data block to the cloud service platform through a high-bandwidth channel.

2. The action camera real-time data transmission and cloud storage management system according to claim 1, characterized in that, The semantic encoding module includes: The data acquisition unit is used to acquire the video stream from the motion camera, the data stream from the inertial measurement unit, and the audio stream. The multimodal fusion analysis unit is used to take the video stream, inertial measurement unit data stream and audio stream as inputs and process them in real time through a preset edge AI coprocessor; The event generation unit is used to generate the semantic event when the multimodal fusion analysis unit identifies a preset action event, visual event, or audio event. The semantic event contains a precise timestamp of the event occurrence point.

3. The action camera real-time data transmission and cloud storage management system according to claim 1, characterized in that, The first communication module includes: The channel selection unit is used to select a low-power, low-bandwidth heartbeat channel as the data transmission carrier. An event reporting unit is used to send the semantic metadata stream, which is encapsulated in the semantic event, to the cloud service platform through the heartbeat channel; The instruction receiving unit is used to receive the data extraction instruction from the cloud service platform through the heartbeat channel.

4. The action camera real-time data transmission and cloud storage management system according to claim 3, characterized in that, The instruction receiving unit is specifically used to receive the data extraction instruction through the heartbeat channel. After the data extraction instruction is generated, it is directly transmitted through the heartbeat channel to ensure that the motion camera can still receive the instruction when the second communication module is not in an active state.

5. The action camera real-time data transmission and cloud storage management system according to claim 1, characterized in that, The data extraction decision module includes: A semantic parsing unit is used to receive the semantic event and parse its attribute information into a queryable data structure. The rule matching unit is used to determine whether the semantic event satisfies the preset semantic subscription rule, wherein the semantic subscription rule is composed of one or more matching predicate logical conjunctions related to the event attributes; The trigger determination unit is used to perform calculations on the determination result of the rule matching unit and the viewing request from the user terminal to generate the data extraction trigger signal.

6. The action camera real-time data transmission and cloud storage management system according to claim 5, characterized in that, The trigger determination unit is specifically used to perform calculations on the determination result of the rule matching unit and the determination result of the viewing request from the user terminal. When the semantic subscription rule is matched or the viewing request is received, the data extraction trigger signal is generated.

7. The action camera real-time data transmission and cloud storage management system according to claim 1, characterized in that, The reverse scheduling module includes: The instruction parameter calculation unit is used to calculate the start timestamp and duration of the video data block required by the data extraction instruction based on the timestamp contained in the semantic event, combined with the preset advance buffer and lag buffer. The instruction generation unit is used to encapsulate the start timestamp, the duration, the unique identifier of the target device, and the data quality specifications into the data extraction instruction; The instruction issuing unit is used to send the data extraction instruction to the motion camera terminal through the first communication module.

8. The action camera real-time data transmission and cloud storage management system according to claim 7, characterized in that, The instruction parameter calculation unit is specifically used to calculate the start timestamp by subtracting a preset advance buffer amount from the timestamp of the semantic event, and to calculate the duration by adding the preset advance buffer amount and the preset lag buffer amount to the duration of the semantic event.

9. The action camera real-time data transmission and cloud storage management system according to claim 1, characterized in that, The second communication module includes: A high-bandwidth channel wake-up unit is used to wake up a high-bandwidth channel that is in a sleep or low-power state by default after receiving the data extraction instruction. The data positioning and reading unit is used to accurately locate and read the video data block from the local storage medium according to the start timestamp and duration contained in the data extraction instruction; The high-speed upload unit is used to upload the video data block to the cloud service platform through the high-bandwidth channel, and to return the high-bandwidth channel to a sleep or low-power state after the upload is completed.

10. A method for real-time data transmission and cloud storage management of an action camera, as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. The semantic encoding module processes real-time sensor data to generate semantic events containing timestamps. S2. The semantic event is sent to the cloud service platform through the first communication module, and the data extraction decision module generates a data extraction trigger signal based on the received semantic event and preset conditions. S3. The reverse scheduling module responds to the data extraction trigger signal to generate a data extraction instruction containing information associated with the timestamp, and sends the data extraction instruction to the motion camera through the first communication module. S4. After the action camera receives the data extraction instruction, the second communication module obtains the corresponding video data block from the local storage medium and uploads the video data block to the cloud service platform through a high-bandwidth channel.

Images