Acquisition strategy adjusting method of low-power-consumption camera and related equipment

Abstract

The invention discloses an acquisition strategy adjustment method of a low-power-consumption camera and related equipment, and the method comprises the steps: waking up a camera main system to carry out high-quality image acquisition on a specified event, and controlling the main system to enter a dormant state and the camera to enter a low-power-consumption mode if the specified event is detected to be ended; acquiring an acquisition strategy parameter initially used for image acquisition in the low power consumption mode, and acquiring a residual power consumption budget of the camera; acquiring a real-time image frame based on the acquisition strategy parameter, and identifying a target object attribute and / or an event type corresponding to the real-time image frame to obtain an image identification result; and adjusting the acquisition strategy parameters based on the image recognition result and the residual power consumption budget. According to the method, event perception omission caused by fixed acquisition strategy parameters can be avoided on the premise of prolonging the standby time of the camera, so that complete capture of key information is ensured to the greatest extent under strict power consumption constraints, and the real-time perception, intelligent decision and dynamic self-adaption capabilities of the low-power-consumption camera are improved.

Description

Technical Field

[0001] This application relates to the field of camera control technology, and in particular to a method for adjusting the acquisition strategy of a low-power camera and related equipment. Background Technology

[0002] To extend the battery life of cameras, low-power cameras typically adopt an event-triggered operating mode. When an event such as motion is detected, the main system is woken up to record high-definition video continuously. After the event ends, a fixed event trigger interval (e.g., 60 seconds) is entered. During this interval, the main system is completely in sleep mode and no longer responds to any events, thus saving power.

[0003] The above-mentioned fixed event triggering interval method has drawbacks such as the risk of missing event detection and inflexible resource allocation. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides at least one method for adjusting the acquisition strategy of a low-power camera and related equipment.

[0005] The first aspect of this application provides a method for adjusting the acquisition strategy of a low-power camera. The method includes: if a specified event is detected, waking up the camera's main system to perform high-quality image acquisition; if the specified event ends, controlling the main system to enter a sleep state and the camera to enter a low-power mode; obtaining the acquisition strategy parameters initially used for image acquisition in the low-power mode, and obtaining the camera's remaining power budget; acquiring real-time image frames based on the acquisition strategy parameters, identifying the target object attributes and / or event types corresponding to the real-time image frames, and obtaining image recognition results; and adjusting the acquisition strategy parameters based on the image recognition results and the remaining power budget.

[0006] In one embodiment, if a specified event is detected, the camera's main system is woken up to perform high-quality image acquisition, including: acquiring image recognition results obtained in low-power mode; if the image recognition results reflect the existence of the specified event, verifying the specified event; if the verification is successful, forcibly waking up the main system to perform high-quality image acquisition.

[0007] In one embodiment, verifying a specified event includes: verifying the specified event using one or more of the following verification rules: Verification rule 1: The specified event is detected in multiple consecutive image frames; Verification rule 2: The specified event is detected by other types of sensors; Verification rule 3: The context information of the specified event meets preset conditions.

[0008] In one embodiment, obtaining the acquisition strategy parameters initially used for image acquisition in low-power mode includes: obtaining event information corresponding to a specified event that has ended; querying the acquisition strategy parameters that match the event information to obtain the acquisition strategy parameters initially used for image acquisition in the current low-power mode.

[0009] In one embodiment, obtaining event information corresponding to the end of a specified event includes: obtaining the target object type detected near the end of the specified event to obtain event information; querying the acquisition strategy parameters that match the event information to obtain the acquisition strategy parameters used for initial image acquisition in the current low-power mode, including: querying the image frame acquisition interval and image acquisition duration that match the target object type to obtain the acquisition strategy parameters used for initial image acquisition in the low-power mode.

[0010] In one embodiment, obtaining the remaining power consumption budget of the camera includes: dividing the data acquisition time period into multiple acquisition time periods and pre-allocating a corresponding power consumption budget for each acquisition time period; calculating the difference between the power consumption budget corresponding to the previous acquisition period and the power consumption already used to obtain the previous power consumption difference; adjusting the power consumption budget corresponding to the current acquisition period based on the previous power consumption difference to obtain the actual power consumption budget of the current acquisition period; wherein, the current acquisition period refers to the acquisition time period at the current moment, and the previous acquisition period refers to the acquisition time period before the current acquisition period; calculating the difference between the actual power consumption budget corresponding to the current acquisition period and the power consumption already used to obtain the remaining power consumption budget of the camera in the current acquisition period.

[0011] In one embodiment, image acquisition based on acquisition strategy parameters yields real-time image frames. Identifying the target object attributes and / or event types corresponding to the real-time image frames to obtain image recognition results includes: detecting the presence of a target in the sensing area of ​​the deployed camera based on a target perception sensor; if a target is detected in the sensing area, acquiring real-time image frames based on acquisition strategy parameters, performing target detection on the real-time image frames to obtain target detection results; if the target detection results are not empty, identifying target object attributes and / or event types for the target objects detected in the real-time image frames to obtain image recognition results.

[0012] In one embodiment, the acquisition strategy parameters include image frame acquisition intervals; based on the image recognition results and the remaining power budget, the acquisition strategy parameters are adjusted, including: querying the image frame acquisition interval that matches the image recognition results to obtain the matching frame interval; if the remaining power budget is lower than a preset power threshold, the matching frame interval is increased to obtain an updated frame interval, and the updated frame interval is used as the adjustment result; if the remaining power budget is not lower than the preset power threshold, the matching frame interval is directly used as the adjustment result.

[0013] A second aspect of this application provides a low-power camera acquisition strategy adjustment device, comprising: an event response module, configured to wake up the camera's main system to perform high-quality image acquisition if a specified event is detected, and to control the main system to enter a sleep state and the camera to enter a low-power mode if the specified event ends; a parameter acquisition module, configured to acquire the acquisition strategy parameters initially used for image acquisition in low-power mode, and to acquire the camera's remaining power budget; an image recognition module, configured to acquire real-time image frames based on the acquisition strategy parameters, identify the target object attributes and / or event types corresponding to the real-time image frames, and obtain image recognition results; and a strategy adjustment module, configured to adjust the acquisition strategy parameters based on the image recognition results and the remaining power budget.

[0014] A third aspect of this application provides a low-power camera, including a memory and a processor, wherein the processor is used to execute program instructions stored in the memory to implement the acquisition strategy adjustment method of the low-power camera described above.

[0015] The fourth aspect of this application provides a computer-readable storage medium storing program instructions thereon, which, when executed by a processor, implement the aforementioned method for adjusting the acquisition strategy of a low-power camera.

[0016] The above scheme involves waking up the camera's main system to acquire high-quality images of a specified event. If the specified event ends, the main system is put into sleep mode, and the camera enters low-power mode. The scheme obtains the initial acquisition strategy parameters used for image acquisition in low-power mode and the camera's remaining power budget. Based on the acquisition strategy parameters, real-time image frames are acquired, and the target object attributes and / or event types corresponding to the real-time image frames are identified to obtain image recognition results. Based on the image recognition results and the remaining power budget, the acquisition strategy parameters are adjusted. This extends the camera's standby time while avoiding event perception omissions caused by fixed acquisition strategy parameters. Thus, under strict power constraints, it maximizes the complete capture of key information, improving the low-power camera's real-time perception, intelligent decision-making, and dynamic adaptation capabilities.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this application. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with this application and, together with the specification, serve to explain the technical solutions of this application.

[0019] Figure 1 This is a flowchart illustrating a method for adjusting the acquisition strategy of a low-power camera, as shown in an exemplary embodiment of this application; Figure 2 This is a schematic diagram illustrating the operation of a low-power camera as shown in an exemplary embodiment of this application; Figure 3 This is a schematic diagram illustrating event verification in an exemplary embodiment of this application; Figure 4 This is a block diagram illustrating a low-power camera acquisition strategy adjustment device according to an exemplary embodiment of this application; Figure 5 This is a schematic diagram of the structure of a low-power camera shown in an exemplary embodiment of this application; Figure 6 This is a schematic diagram illustrating the structure of a computer-readable storage medium, as shown in an exemplary embodiment of this application. Detailed Implementation

[0020] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0021] In the following description, specific details such as particular system architectures, interfaces, and technologies are presented for illustrative purposes rather than for limiting purposes, in order to provide a thorough understanding of this application.

[0022] In this document, the term "and / or" is merely a description of the association information of related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, "many" in this document means two or more. Moreover, the term "at least one" in this document means any combination of at least two of any one or more of a plurality of elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0023] The following describes the method for adjusting the acquisition strategy of a low-power camera provided in the embodiments of this application.

[0024] The low-power camera acquisition strategy adjustment method provided in this application embodiment can be executed by the low-power camera, other computing devices that are communicatively connected to the low-power camera, or by the low-power camera and other computing devices interacting and cooperating. This application does not limit this.

[0025] Please see Figure 1 , Figure 1 This is a flowchart illustrating an exemplary embodiment of the acquisition strategy adjustment method for a low-power camera, as shown in this application. Figure 1 As shown, the method for adjusting the acquisition strategy of a low-power camera includes at least steps S110 to S140, which are described in detail below: Step S110: If a specified event is detected, the camera's main system is woken up to acquire high-quality images. If the specified event ends, the main system is controlled to enter a sleep state and the camera enters a low-power mode.

[0026] The main system refers to the set of core computing units inside the camera that are responsible for handling complex, high-power tasks, such as high-definition video encoders and high-performance artificial intelligence (AI) processors.

[0027] Among them, the low-power mode can be the Always On Vision (AOV) recording mode.

[0028] In low-power mode, the camera's main system enters a sleep state to save power.

[0029] For example, in AOV mode, most of the camera hardware (such as the main processor, high-resolution image sensor, etc.) is powered off or in sleep mode. An independent, low-power coprocessor (such as a micro controller unit (MCU) or a low-power video processing unit (VPU)) continues to work. It is connected to a low-power image sensor and performs simple analysis tasks (such as pixel change detection, specific shape recognition, etc.). The power consumption used during operation is much lower than the power consumption used by the main system during execution.

[0030] In low-power mode, low-resolution and / or low-frame-rate image data is analyzed to determine if a specified event exists. If the specified event exists, the low-power mode is interrupted, and the camera's main system is woken up to acquire high-quality images.

[0031] For example, in AOV mode, the coprocessor continuously analyzes low-resolution and / or low-frame-rate image data. Once its built-in lightweight algorithm detects a specified event (such as movement in the frame, the appearance of a target object of a specified shape, the appearance of a vehicle, etc.), it triggers the generation of a main system wake-up signal to wake up the camera's main system to perform high-quality image acquisition for the specified event.

[0032] If the specified event is detected to have ended, the main system will re-enter hibernation mode, and the camera will re-enter low-power mode to save power.

[0033] Step S120: Obtain the acquisition strategy parameters initially used for image acquisition in low-power mode, and obtain the remaining power budget of the camera.

[0034] After entering low-power mode, obtain the acquisition strategy parameters initially used for image acquisition in the current low-power mode.

[0035] The acquisition strategy parameters specifically refer to the parameters that the camera can adjust and that affect power consumption when acquiring images, including but not limited to image frame acquisition interval, and / or image frame resolution, and / or duration, etc. This application does not limit these parameters.

[0036] For example, the initial acquisition strategy parameters can be pre-set; or the initial acquisition strategy parameters adapted to the current low-power mode can be flexibly selected according to the actual situation. This application does not limit this.

[0037] Additionally, obtain the camera's remaining power consumption budget.

[0038] Specifically, a power consumption budget is pre-allocated, and the remaining power consumption budget is obtained based on the allocated power consumption budget and the power consumption budget that has already been used.

[0039] For example, the data acquisition period can be divided according to time, such as one hour, one day, one week, or one month, resulting in multiple acquisition time periods. A corresponding power budget can be allocated to each acquisition time period to balance power distribution across different times and ensure long-term sustainable operation of the camera. Then, based on the power budget already used within the current acquisition time period, the remaining power budget within the current acquisition time period can be calculated to obtain the remaining power budget.

[0040] For example, power budgets can be allocated to different image acquisition modes of the camera, such as allocating power budgets to low-power mode and high-quality image acquisition mode of the main system, respectively. This allows for power allocation based on the characteristics of different image acquisition modes. By adjusting different power allocation ratios, it can be determined whether the camera is geared towards long-term continuous operation or high-quality event perception. Then, based on the power budget already used in low-power mode, the remaining power budget for the current low-power mode can be calculated to obtain the remaining power budget.

[0041] For example, first, an initial power budget is allocated to each acquisition time period. Then, based on the initial power budget allocated to each acquisition time period, a corresponding power budget is allocated to different image acquisition modes within that acquisition time period. For instance, the initial power budget allocated to acquisition time period 'a' is 5% of the battery. Further, a power budget of 1% of the battery is allocated to the low-power mode within acquisition time period 'a', and a power budget of 4% is allocated to the high-quality image acquisition mode of the main system. Then, based on the power budget already used by the low-power mode within the current acquisition time period, the remaining power budget for the current low-power mode is calculated to obtain the remaining power budget.

[0042] The specific method for allocating power consumption budget can be set by the user or flexibly set according to the specific application scenario.

[0043] For example, obtain the remaining power of the camera, determine the power budget allocation method based on the remaining power, allocate the power budget according to the power budget allocation method, obtain the allocated power budget, calculate the usable power budget based on the allocated power budget and the used power budget, and obtain the remaining power budget of the camera.

[0044] For example, if the camera's remaining power is less than 80% but greater than 40%, a corresponding power budget is allocated according to the acquisition time period to ensure event perception capability at each time when the power is sufficient, and to control the camera's continuous operating time. If the camera's remaining power is less than 40% but greater than 20%, an initial power budget is allocated according to the acquisition time period, and then a corresponding power budget is allocated to different image acquisition modes within that acquisition time period based on the initial power budget to avoid the main system missing the recording of important events. If the camera's remaining power is less than 20%, corresponding power budgets are allocated to the low-power mode and the main system's high-quality image acquisition mode respectively to extend the camera's continuous operating time as much as possible.

[0045] It should be noted that, in the various power consumption budget allocation methods exemplified above, the specific values ​​of the power consumption budgets allocated for different acquisition time periods and / or different image acquisition modes can be determined based on the power consumption usage of different acquisition time periods and / or different image acquisition modes within a historical time period, or they can be set by the user. This application does not impose any limitations on this.

[0046] Step S130: Based on the acquisition strategy parameters, acquire images to obtain real-time image frames, identify the target object attributes and / or event types corresponding to the real-time image frames, and obtain image recognition results.

[0047] Based on the acquisition strategy parameters, real-time image frames are obtained through image acquisition.

[0048] For example, the image frame acquisition interval, and / or image frame resolution, and / or duration are determined based on the acquisition strategy parameters in order to acquire real-time image frames.

[0049] Identify the target object attributes and / or event types in real-time image frames to obtain image recognition results.

[0050] The target object attributes include, but are not limited to, one or more of the following: target object type, target object posture, target object position, etc. The event type can be such as weather, target object behavior, target object action, etc. This application does not limit the target object attributes and event types.

[0051] It should be noted that, in order to reduce power consumption, a lightweight algorithm is used to analyze real-time image frames.

[0052] Step S140: Adjust the acquisition strategy parameters based on the image recognition results and the remaining power consumption budget.

[0053] The acquisition strategy parameters are adjusted based on the image recognition results and the remaining power consumption budget.

[0054] For example, taking the adjustment of the image frame acquisition interval as an example, the image frame acquisition interval in the acquisition strategy parameters before the adjustment is 5 seconds. The camera acquires a low-resolution image every 5 seconds, and the low-resolution image is identified to obtain the image recognition result. If the current image recognition result indicates the existence of a specified target object, the image frame acquisition interval corresponding to the specified target object is obtained at 2 seconds. If the remaining power budget is greater than the preset threshold, the camera's image frame acquisition interval is updated to 2 seconds; if the remaining power budget is less than the preset threshold, the camera's image frame acquisition interval is updated to 4 seconds.

[0055] Of course, in addition to adjusting the image frame acquisition interval, other acquisition strategy parameters, such as image frame resolution and / or duration, can also be adjusted, and this application does not limit these parameters.

[0056] In addition, the acquisition strategy parameters in low-power mode are adjusted in real time. That is, as soon as a new image recognition result is detected, the acquisition strategy parameters are adjusted accordingly to obtain acquisition strategy parameters that are more suitable for the actual scenario.

[0057] For example, please see Figure 2 , Figure 2 This is a schematic diagram illustrating the operation of a low-power camera as shown in an exemplary embodiment of this application, such as... Figure 2 As shown, if a specified event is detected in any mode, the main system is awakened to record. If the specified event ends, the system enters AOV mode. In AOV mode, real-time image frames are acquired at the image frame acquisition interval corresponding to the acquisition strategy parameters. The real-time image frames are identified, and if a new target object exists, the image frame acquisition interval is adjusted according to the target object type. If a specified event exists, the main system is awakened to record. If the duration of AOV mode reaches the image frame acquisition duration corresponding to the acquisition strategy parameters, the AOV mode duration ends, and the camera enters deep sleep mode. The power consumption of deep sleep mode is lower than that of low-power mode. For example, the image frame acquisition interval is the largest in deep sleep mode, or image acquisition is not performed in deep sleep mode, only the LiDAR sensor, infrared sensor, etc., are retained for environmental perception.

[0058] This application combines image recognition results and remaining power budget to adjust the acquisition strategy parameters of the low-power mode in a timely and flexible manner. While extending the camera standby time, it avoids missing event perception caused by fixed acquisition strategy parameters. Thus, under strict power constraints, it maximizes the complete capture of key information and improves the real-time perception, intelligent decision-making and dynamic adaptation capabilities of the low-power mode.

[0059] The following describes some embodiments of this application in detail.

[0060] In some implementations, if a specified event is detected in step S110, the camera's main system is woken up to perform high-quality image acquisition, including the following steps S111 to S112.

[0061] Step S111: Obtain the image recognition result obtained in low power mode. If the image recognition result reflects the existence of a specified event, then verify the specified event.

[0062] A list of specified events can be pre-set, which stores a list of high-priority event types. If the event type corresponding to the image recognition result is in the specified event list, it is determined that the specified event exists.

[0063] It can also be the event type corresponding to the image recognition result, including the event risk level. If the event risk level is higher than the preset risk level, it is determined that the specified event exists.

[0064] If the image recognition result indicates the existence of a specified event, the main system is not directly woken up. Instead, the specified event is verified to improve reliability.

[0065] For example, use one or more of the following validation rules to validate a specified event: Verification rule 1: The specified event is detected in multiple consecutive image frames; Verification Rule 2: Other types of sensors detect the specified event; Verification rule 3: The context information of the specified event meets the preset conditions.

[0066] For verification rule one, if the same specified event is detected in multiple consecutive image frames, the verification is successful.

[0067] Regarding verification rule two, other types of sensors include, but are not limited to, lidar sensors and / or infrared sensors, etc., and this application does not limit them. If other types of sensors also detect the existence of the specified event, the verification passes. For example, if the specified event detected by the image recognition result is the presence of a target object entering a preset area, and if the lidar sensor detects a moving target object within the preset area, or the infrared sensor detects a heat source within the preset area, the verification passes.

[0068] For verification rule three, contextual information includes, but is not limited to, location information, and / or time information, and / or environmental information. Verification passes if the contextual information of the specified event meets preset conditions. For example, if the image recognition result detects a specified event as the presence of a target object entering a preset area, and the target object's location information indicates that its trajectory is unreasonable (e.g., a sudden change in location, or being in an area the target object could not possibly reach), verification fails. Alternatively, if the specified event occurs within a preset time period, it is normal, and verification fails. If the specified event occurs outside the preset time period, it is abnormal, and verification passes. Or, if severe weather causes image degradation, the confidence threshold needs to be increased. If the confidence of the specified event is still greater than the increased confidence threshold, verification passes.

[0069] Step S112: If the verification is successful, the main system will be forcibly woken up to perform high-quality image acquisition.

[0070] If the specified event passes one or more of the above verification rules, the current specified event is considered trustworthy, and the main system is forcibly woken up to perform high-quality image acquisition. Power consumption is not considered when forcibly waking up.

[0071] For example, please see Figure 3 , Figure 3 This is a schematic diagram illustrating event verification as shown in an exemplary embodiment of this application, such as... Figure 3 As shown, the image recognition results indicate the existence of a suspicious specified event. Verification rule one is used for time-series verification, verification rule two is used for sensor verification, and verification rule three is used for context verification to determine whether the verification passes. If all the above multi-level verifications pass, the main system is forcibly woken up to perform high-quality image acquisition. If one level of verification fails, the low-power mode is maintained.

[0072] Optionally, if a specified event is detected and the main system is woken up, the main system's high-precision recognition model can also detect whether the specified event exists. If it exists, high-quality image acquisition is performed; otherwise, the main system is controlled to re-enter sleep mode and the camera re-enters low-power mode.

[0073] If a specified event is detected, the camera's main system is woken up to acquire high-quality images, ensuring that images or videos of the specified event of better quality can be acquired. This continues until the specified event is detected to have ended. Then, the main system is controlled to re-enter sleep mode, and the camera re-enters low-power mode. The camera then obtains the initial acquisition strategy parameters used for image acquisition in the current low-power mode and performs image acquisition in low-power mode.

[0074] In some implementations, step S120 involves obtaining the acquisition strategy parameters initially used for image acquisition in low-power mode, including steps S1211 to S1212 below.

[0075] Step S1211: Obtain the event information corresponding to the specified event that has ended.

[0076] The event information includes, but is not limited to, the type or number of target objects detected during the duration of the event, the type of target objects detected near the end of the event, the duration of the event, or the risk level of the event. This application does not limit the specific content of the event information.

[0077] Step S1212: Query the acquisition strategy parameters that match the event information to obtain the acquisition strategy parameters used initially for image acquisition in the current low-power mode.

[0078] There is a pre-set mapping relationship between different event information and different acquisition strategy parameters. Based on the current event information, the matching acquisition strategy parameters are queried and used as the initial acquisition strategy parameters for image acquisition in the current low power mode.

[0079] For example, the target object type detected near the end of a specified event is obtained to get the event information; the image frame acquisition interval and image acquisition duration matching the target object type are queried to obtain the acquisition strategy parameters used for initial image acquisition in low-power mode.

[0080] In addition, it is necessary to obtain the camera's remaining power consumption budget in order to adjust the acquisition strategy parameters later.

[0081] A specific embodiment will be described in detail. Step S120 involves obtaining the remaining power consumption budget of the camera, including steps S1221 to S1224.

[0082] Step S1221: Divide the data acquisition time period into multiple periods and pre-allocate the corresponding power consumption budget for each data acquisition time period.

[0083] The data collection time period is divided into multiple time periods, such as hours, days, and weeks, and a corresponding power consumption budget is pre-allocated for each time period.

[0084] The power consumption budget can be allocated equally for each acquisition time period, in which case the power consumption budget for each acquisition time period is the same. Of course, the power consumption budget can also be allocated non-equally for each acquisition time period, in which case the power consumption budget for each acquisition time period is different.

[0085] For example, the power consumption of each acquisition time period and / or the frequency of a specified event are statistically analyzed within a historical time period to obtain historical characteristics; based on the historical characteristics, a power consumption budget is allocated for each acquisition time period to obtain the power consumption budget corresponding to each acquisition time period.

[0086] Specifically, if the power consumption of the same acquisition time period within a historical time period is higher and / or the frequency of the specified event is higher, the power consumption budget allocated to that acquisition time period will be higher; conversely, if the power consumption of the same acquisition time period within a historical time period is lower and / or the frequency of the specified event is lower, the power consumption budget allocated to that acquisition time period will be lower.

[0087] For example, the camera deployment location is obtained, and the event perception level of the deployment scene is determined based on the camera deployment location. If the event perception level is higher, a higher power consumption budget is allocated for each acquisition time period; if the event perception level is lower, a lower power consumption budget is allocated for each acquisition time period.

[0088] For example, if the camera's Global Positioning System (GPS) information shows that the camera is located in a "villa courtyard" with a high event awareness level, the system automatically adopts the high standard power budget configured for that area, allowing the camera to perform more frequent high frame rate image acquisition in low power mode.

[0089] The above embodiments allow for flexible allocation of power consumption budgets, improving the accuracy of power consumption budget allocation.

[0090] Step S1222: Calculate the difference between the power budget corresponding to the preceding acquisition cycle and the power consumption already used to obtain the preceding power consumption difference.

[0091] The preceding acquisition period refers to the acquisition time period before the current acquisition period, while the current acquisition period refers to the acquisition time period at which the current moment occurs.

[0092] Obtain the power budget corresponding to the preceding acquisition cycle and the power consumption used in the preceding acquisition cycle. Calculate the difference between the two to obtain the preceding power consumption difference.

[0093] The preceding acquisition period used to calculate the difference in preceding power consumption can be one acquisition time period before the current acquisition period, or multiple acquisition time periods before the current acquisition period. This application does not limit this.

[0094] Step S1223: Based on the power consumption difference of the previous period, adjust the power consumption budget corresponding to the current acquisition cycle to obtain the actual power consumption budget of the current acquisition cycle.

[0095] The power budget for the current acquisition cycle is adjusted based on the power difference from the previous acquisition cycle to obtain the actual power budget for the current acquisition cycle.

[0096] Specifically, if the power consumption difference of the preceding period is less than 0, it indicates that there is a power consumption overflow in the preceding acquisition period. In this case, the power consumption budget corresponding to the current acquisition period needs to be reduced accordingly. If the power consumption difference of the preceding period is less than 0, it indicates that there is a power consumption budget surplus in the preceding acquisition period. In this case, the power consumption budget corresponding to the current acquisition period can be increased accordingly, or the initial power consumption budget allocated in the current acquisition period can be directly used as the final actual power consumption budget.

[0097] By adjusting the power budget for the current acquisition cycle based on the actual power consumption of the previous acquisition cycle, power resource management can be carried out in conjunction with multiple acquisition time cycles, thereby improving the balance and stability of power consumption among the acquisition time cycles.

[0098] Step S1224: Calculate the difference between the actual power consumption budget and the power consumption already used for the current acquisition cycle to obtain the remaining power consumption budget of the camera in the current acquisition cycle.

[0099] Obtain the power consumption used in the current acquisition cycle, and subtract it from the actual power consumption budget to obtain the remaining power consumption budget of the camera in the current acquisition cycle.

[0100] In addition, real-time image frames are obtained by acquiring images according to the acquisition strategy parameters, and image recognition results are obtained by recognizing the real-time image frames.

[0101] In some implementations, step S130 involves acquiring real-time image frames based on acquisition strategy parameters, identifying the target object attributes and / or event types corresponding to the real-time image frames, and obtaining image recognition results, including steps S131 to S133.

[0102] Step S131: Detect the presence of a target in the sensing area of ​​the deployed camera based on the target perception sensor.

[0103] Target perception sensors are used to detect the presence of targets within a perception area. Target perception sensors include, but are not limited to, lidar sensors and / or infrared sensors.

[0104] For example, if the infrared sensor detects a change in heat source in the sensing area, and / or if the lidar sensor detects a moving object in the sensing area, then it is determined that a target exists in the sensing area; otherwise, it is determined that no target exists in the sensing area.

[0105] Step S132: If a target is detected in the perception area, image acquisition is performed based on the acquisition strategy parameters to obtain a real-time image frame, and target detection is performed on the real-time image frame to obtain the target detection result.

[0106] If the target perception sensor detects a target in the perception area, it performs target detection on the real-time image frames acquired based on the acquisition strategy parameters to obtain the target detection result.

[0107] For example, target bounding boxes are marked on target objects in real-time image frames to obtain target detection results.

[0108] Step S133: If the target detection result is not empty, then target object attribute recognition and / or event type recognition are performed on the target object detected in the real-time image frame to obtain the image recognition result.

[0109] If the target detection result is not empty (e.g., the target box is marked in the real-time image frame), then target object attribute recognition and / or event type recognition are performed on the target objects detected in the real-time image frame. For example, the target objects in the target box are classified and / or events are recognized to obtain the image recognition result.

[0110] The above embodiments divide the image recognition process into multi-level recognition, and the next level is only started after the previous level is passed, in order to save power consumption.

[0111] The acquisition strategy parameters are adjusted based on the image recognition results and the remaining power consumption budget.

[0112] In some implementations, the acquisition strategy parameters include the image frame acquisition interval; in step S140, the acquisition strategy parameters are adjusted based on the image recognition results and the remaining power consumption budget, including steps S141 to S143.

[0113] Step S141: Query the image frame acquisition interval that matches the image recognition result to obtain the matching frame interval.

[0114] A mapping relationship is constructed between different image recognition results and different image frame acquisition intervals. For example, a strategy mapping table is generated by predefining or learning from sample data. By querying the strategy mapping table, the image frame acquisition interval that matches the image recognition result is obtained, thus obtaining the matching frame interval.

[0115] Step S142: If the remaining power consumption budget is lower than the preset power consumption threshold, the matching frame interval is increased to obtain the updated frame interval, and the updated frame interval is used as the adjustment result.

[0116] If the remaining power budget is lower than the preset power threshold, it indicates that the current remaining power budget is insufficient. In this case, the strategy is downgraded, the matching frame interval is increased, and the updated frame interval is used as the adjustment result for subsequent image acquisition to reduce power consumption.

[0117] Step S143: If the remaining power consumption budget is not lower than the preset power consumption threshold, the matching frame interval is directly used as the adjustment result.

[0118] If the remaining power budget is not lower than the preset power threshold, it indicates that the current remaining power budget is sufficient. At this time, the original strategy can be maintained, and the matching frame interval can be directly used as the adjustment result for subsequent image acquisition.

[0119] Of course, when the remaining power budget is not lower than the preset power threshold, the matching frame interval can be reduced to allow for more frequent image acquisition and improve event awareness.

[0120] The aforementioned preset power consumption threshold can be flexibly set according to the actual situation. For example, the preset power consumption threshold can be flexibly set according to the current time and the current stage of the current acquisition cycle (such as the early stage, middle stage, and late stage) to ensure that the power consumption is correctly judged.

[0121] In some implementations, the decision to wake up the main system can be determined based on the remaining power budget. For example, it can be checked whether the remaining power budget is lower than the minimum power threshold. If so, the main system will not be woken up. In this way, even if a specified event is detected in low power mode, the main system will not be woken up to ensure the camera's battery life.

[0122] In some implementations, the acquisition strategy parameters can be adjusted by combining more other sensors, such as infrared sensors, and / or lidar sensors, and / or GPS.

[0123] For example, infrared sensors and / or lidar sensors act as "sentinels," continuously sensing the environment with extremely low power consumption. Their trigger signals can serve as conditions for initiating image acquisition or increasing the image acquisition frame rate in low-power mode. If the camera is in deep sleep mode, and the infrared sensor and / or lidar sensor detects motion, it enters low-power mode and performs image acquisition based on the acquisition strategy parameters of low-power mode. Alternatively, if the camera is in low-power mode and the infrared sensor and / or lidar sensor detects motion, it first increases the image acquisition frame rate, performs image acquisition based on the increased frame rate, and then adjusts the acquisition strategy parameters based on the image recognition results and the remaining power budget.

[0124] In addition, when performing specified event detection, the confidence level of the specified event can be determined by combining image recognition results, infrared sensor and lidar sensor.

[0125] Furthermore, based on the geographic location context provided by GPS or BeiDou information, different acquisition strategy parameters can be adapted for cameras in different locations, such as adaptively switching acquisition strategy parameters according to different areas such as "home" and "office".

[0126] In some implementations, a status watchdog can be set. A status watchdog is a mechanism used in embedded systems to monitor the running status of programs. It prevents the system from freezing or crashing by resetting periodically. If the status watchdog times out, the camera is reset to a safe baseline.

[0127] In some implementations, unified time stamp management is applied to image sequences acquired in normal recording, designated event recording, and low-power mode to generate logs with fused timelines and event summaries, facilitating the analysis of camera historical operation and improving user experience.

[0128] The following example illustrates this with a specific application scenario: For example, an AOV camera includes a main processor, a low-power coprocessor, an image sensor (for image acquisition), a memory (for storing computer programs), a communication module (for data communication), and peripheral sensors (such as infrared sensors, and / or lidar sensors, and / or GPS). The low-power coprocessor has an intelligent decision engine deployed in it, which is used to adjust the acquisition strategy parameters adopted in low-power mode.

[0129] Specifically, the camera's main processor triggers recording when it detects a specified event. When the specified event is about to end, the detected target object type is Type 1. The specified event ends, the main processor goes into sleep mode, and enters AOV mode.

[0130] The low-power coprocessor's intelligent decision engine starts, queries the policy mapping table, and sets the initial image acquisition interval to 2 seconds and the duration to 30 seconds according to "Type 1".

[0131] In AOV mode, the coprocessor operates at 2 seconds per frame. At the 10th second, a new target object type is identified as "Type II". The intelligent decision engine queries the strategy mapping table and immediately switches the image acquisition interval to 5 seconds, which corresponds to "Type II".

[0132] At the 15-second mark, a specified event is detected, and event verification begins. If the specified event is detected in three consecutive frames, the main processor is forcibly woken up. After confirming the existence of the specified event, the main processor executes high-definition recording and generates alarm information.

[0133] If no event is specified, the system will enter baseline AOV mode after 30 seconds. The image acquisition interval in baseline AOV mode is 10 seconds.

[0134] In addition, the data acquisition strategy can be adjusted by combining multiple sensors: For example, when the camera is in deep sleep mode, if the infrared sensor detects a change in heat source, it wakes up the low-power coprocessor and activates from the "no acquisition" state of deep sleep mode to the "10 seconds / frame" state of baseline AOV mode. In baseline AOV mode, if the radar detects a target moving rapidly towards the camera, the intelligent decision engine immediately switches the image acquisition interval from 10 seconds to 5 seconds before identifying the specific target based on the image. Subsequently, the image recognition result indicates that the target object type is "Type II". The intelligent decision engine confirms and maintains this policy according to the policy mapping table. Then, the image recognition result indicates the existence of a specified event. The intelligent decision engine immediately queries the LiDAR data and confirms the existence of a stationary or moving target object ahead based on the LiDAR data. The dual confirmation from the image and radar increases the confidence of the specified event, and the main processor is immediately woken up.

[0135] Figure 4 This is a block diagram illustrating a low-power camera acquisition strategy adjustment device, as shown in an exemplary embodiment of this application. Figure 4 As shown, the exemplary low-power camera acquisition strategy adjustment device 400 includes: The event response module 410 is used to wake up the camera's main system to perform high-quality image acquisition if a specified event is detected, and to control the main system to enter a sleep state and the camera to enter a low-power mode if the specified event is detected to have ended. The parameter acquisition module 420 is used to acquire the acquisition strategy parameters used initially for image acquisition in low-power mode, and to acquire the remaining power budget of the camera. Image recognition module 430 is used to acquire real-time image frames based on acquisition strategy parameters, identify the target object attributes and / or event types corresponding to the real-time image frames, and obtain image recognition results. The strategy adjustment module 440 is used to adjust the acquisition strategy parameters based on the image recognition results and the remaining power consumption budget.

[0136] It should be noted that the low-power camera acquisition strategy adjustment device and the low-power camera acquisition strategy adjustment method provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the low-power camera acquisition strategy adjustment device provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation.

[0137] Please see Figure 5 , Figure 5This is a schematic diagram illustrating the structure of a low-power camera according to another exemplary embodiment of this application. The low-power camera includes an image sensor 501, a memory 502, and a processor 503. The image sensor 501 is used for image acquisition, and the processor 503 is used to execute program instructions stored in the memory 502 to implement the steps in any of the above embodiments of the low-power camera acquisition strategy adjustment method. The processor 503 includes a main processor and a low-power coprocessor.

[0138] Specifically, processor 503 controls itself and memory 502 to implement the steps in any of the above-described embodiments of the low-power camera acquisition strategy adjustment method. Processor 503 can also be referred to as a Central Processing Unit (CPU). Processor 503 may be an integrated circuit chip with signal processing capabilities. Processor 503 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor. Furthermore, processor 503 can be implemented using integrated circuit chips.

[0139] Please see Figure 6 , Figure 6 This is a schematic diagram illustrating the structure of a computer-readable storage medium in an exemplary embodiment of this application. The computer-readable storage medium 600 stores program instructions 610 that can be executed by a processor. The program instructions 610 are used to implement the steps in any of the above embodiments of the low-power camera acquisition strategy adjustment method.

[0140] In some embodiments, the functions or modules of the apparatus provided in this disclosure can be used to perform the methods described in the above method embodiments. The specific implementation can be referred to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0141] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0142] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus implementations described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0143] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for adjusting the acquisition strategy of a low-power camera, characterized in that, The method includes: If a specified event is detected, the camera's main system is woken up to acquire high-quality images. If the specified event ends, the main system is controlled to enter a sleep state, and the camera enters a low-power mode. Obtain the acquisition strategy parameters initially used for image acquisition in the low-power mode, and obtain the remaining power budget of the camera; Based on the acquisition strategy parameters, real-time image frames are acquired, and the target object attributes and / or event types corresponding to the real-time image frames are identified to obtain image recognition results. Based on the image recognition results and the remaining power consumption budget, the acquisition strategy parameters are adjusted.

2. The method according to claim 1, characterized in that, If a specified event is detected, the camera's main system is awakened to perform high-quality image acquisition, including: Obtain the image recognition result obtained in low power mode; if the image recognition result reflects the existence of a specified event, then verify the specified event. If the verification passes, the main system will be forcibly woken up to perform high-quality image acquisition.

3. The method according to claim 2, characterized in that, The verification of the specified event includes: The specified event is verified using one or more of the following verification rules: Verification rule 1: The specified event is detected in multiple consecutive image frames; Verification Rule 2: Other types of sensors detect the specified event; Verification rule three: The context information of the specified event meets the preset conditions.

4. The method according to claim 1, characterized in that, The step of obtaining the acquisition strategy parameters initially used for image acquisition in the low-power mode includes: Retrieve event information corresponding to the end of a specified event; The acquisition strategy parameters that match the event information are queried to obtain the acquisition strategy parameters initially used for image acquisition in the current low-power mode.

5. The method according to claim 4, characterized in that, The process of obtaining the event information corresponding to the specified event that has ended includes: Obtain the type of the target object detected near the end of the specified event to obtain event information; The query matches the event information to obtain the acquisition strategy parameters initially used for image acquisition in the current low-power mode, including: The image frame acquisition interval and image acquisition duration that match the target object type are queried to obtain the acquisition strategy parameters used initially for image acquisition in the low-power mode.

6. The method according to claim 1, characterized in that, The process of obtaining the remaining power consumption budget of the camera includes: Divide the data acquisition time period into multiple periods and pre-allocate a corresponding power consumption budget for each period. The difference between the power budget corresponding to the preceding acquisition cycle and the power consumption already used is calculated to obtain the preceding power consumption difference. Based on the preceding power consumption difference, the power consumption budget corresponding to the current acquisition cycle is adjusted to obtain the actual power consumption budget of the current acquisition cycle; wherein, the current acquisition cycle refers to the acquisition time cycle at the current moment, and the preceding acquisition cycle refers to the acquisition time cycle before the current acquisition cycle. Calculate the difference between the actual power consumption budget and the power consumption already used for the current acquisition period to obtain the remaining power consumption budget of the camera in the current acquisition period.

7. The method according to claim 1, characterized in that, The process of acquiring real-time image frames based on the acquisition strategy parameters, identifying the target object attributes and / or event types corresponding to the real-time image frames, and obtaining image recognition results includes: The presence of a target is detected in the sensing area where the camera is deployed, based on the target perception sensor; If a target is detected in the perception area, an image is acquired based on the acquisition strategy parameters to obtain a real-time image frame, and a target detection is performed on the real-time image frame to obtain a target detection result. If the target detection result is not empty, then target object attribute recognition and / or event type recognition are performed on the target object detected in the real-time image frame to obtain the image recognition result.

8. The method according to claim 1, characterized in that, The acquisition strategy parameters include the image frame acquisition interval; adjusting the acquisition strategy parameters based on the image recognition result and the remaining power consumption budget includes: The matching frame interval is obtained by querying the image frame acquisition interval that matches the image recognition result. If the remaining power consumption budget is lower than the preset power consumption threshold, the matching frame interval is increased to obtain the updated frame interval, and the updated frame interval is used as the adjustment result. If the remaining power consumption budget is not lower than the preset power consumption threshold, then the matching frame interval is directly used as the adjustment result.

9. A low-power camera, characterized in that, The low-power camera includes an image sensor, a memory, and a processor. The image sensor is used for image acquisition, and the processor is used to execute program instructions stored in the memory to implement the steps in the method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions that can be executed by a processor to implement the steps of the method as described in any one of claims 1-8.

Images