Target object detection method, device, system, electronic device, and storage medium

Abstract

The application discloses a target object detection method, device and system, electronic equipment and a storage medium. The method comprises the following steps: determining at least one current target object region in a current image, and determining a detection region according to the current target object region; sending the position of the detection region to a collection chip, so that the collection chip performs cropping, scaling and splicing processing on a to-be-processed image collected after the current image according to the position of the detection region; receiving a target image obtained after the collection chip processes the to-be-processed image, and performing target object detection according to the target image. The application can improve the detection effect of the target object under the transmission bandwidth limitation between the collection chip and the system chip.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a target object detection method, apparatus, system, electronic device, and storage medium. Background Technology

[0002] With the development of intelligent chips, more and more DVR (Digital Video Recorder) devices will integrate intelligent functions to achieve tasks such as target object detection. A DVR device consists of an acquisition chip (AD, Analog-to-Digital Converter), a system-on-chip (SOC), and the BT656 transmission protocol. The acquisition chip (AD) acquires images from the connected analog camera and converts them into digital signals in YUV (Luminance and Chrominance) format. These signals are then transmitted to the SOC via the BT656 transmission protocol. The SOC uses the received YUV data to perform intelligent detection, video encoding, and video display functions.

[0003] Due to the bandwidth limitations of the BT656 transmission protocol, the acquisition chip AD typically scales the image before transmission. This results in the system chip SOC receiving a scaled image, leading to lower image clarity and poor target object detection performance when the system chip SOC performs target object detection. Summary of the Invention

[0004] This invention provides a target object detection method, apparatus, system, electronic device, and storage medium to improve the target object detection effect under the limitation of transmission bandwidth between the acquisition chip and the system chip.

[0005] In a first aspect, embodiments of the present invention provide a target object detection method, which is executed by a system chip, and the method includes:

[0006] Identify at least one current target object region in the current image, and determine the detection region based on the current target object region;

[0007] The location of the detection area is sent to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image based on the location of the detection area.

[0008] The receiver chip processes the image to be processed to obtain the target image, and then performs target object detection based on the target image.

[0009] Secondly, embodiments of the present invention also provide another target object detection method, which is executed by a data acquisition chip, and the method includes:

[0010] If the location of the detection area sent by the system chip is determined, the image to be processed is cropped according to the location of the detection area to obtain the detection area image;

[0011] The image to be processed is scaled, and the scaled image to be processed is stitched together with the detection area image to obtain the target image;

[0012] The target image is sent to the system chip so that the system chip can perform target object detection based on the target image.

[0013] Thirdly, embodiments of the present invention also provide a target object detection device, which is deployed on a system-on-a-chip and includes:

[0014] The detection region determination module is used to determine at least one current target object region in the current image, and determine the detection region based on the current target object region;

[0015] The detection area location sending module is used to send the location of the detection area to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image based on the location of the detection area.

[0016] The target object detection module is used to receive the target image obtained after the acquisition chip processes the image to be processed, and to perform target object detection based on the target image.

[0017] Fourthly, embodiments of the present invention also provide another target object detection device, which is deployed on a data acquisition chip, and the device includes:

[0018] The image to be processed cropping module is used to crop the image to be processed according to the detection area position sent by the system chip if the detection area position is determined to be received, so as to obtain the detection area image.

[0019] The image stitching module is used to scale the image to be processed and stitch the scaled image to be processed with the detection area image to obtain the target image.

[0020] The target image sending module is used to send the target image to the system chip so that the system chip can perform target object detection based on the target image.

[0021] Fifthly, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the target object detection method as described in any of the embodiments of the present invention.

[0022] Sixthly, embodiments of the present invention also provide a storage medium for storing computer-executable instructions, which, when executed by a computer processor, are used to perform the target object detection method as described in any of the embodiments of the present invention.

[0023] The technical solution of this invention involves a system chip (SIC) performing target object recognition on the current image sent by the acquisition chip to obtain the current target object region. Based on this region, a detection region is determined, and its location is sent to the acquisition chip. This allows the acquisition chip to crop, scale, and stitch subsequent images based on the detection region location. The processed target image is then sent back to the SIC, which performs target object detection based on this image. This solves the problem in existing technologies where the acquisition chip scales the image before transmission, resulting in low target object clarity and poor detection performance when the SIC performs target object detection on the received image. The technical solution of this invention ensures the clarity of the target object detection region in the image received by the SIC, even under bandwidth limitations between the acquisition chip and the SIC, thus improving the target object detection effect.

[0024] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of a target object detection method provided in Embodiment 1 of the present invention;

[0027] Figure 2 This is a schematic diagram of a detection area provided in Embodiment 1 of the present invention;

[0028] Figure 3 This is a schematic diagram of the structure of a DVR device provided in Embodiment 1 of the present invention;

[0029] Figure 4 This is a flowchart of another target object detection method provided in Embodiment 2 of the present invention;

[0030] Figure 5 This is a schematic diagram of a detection area image provided in Embodiment 2 of the present invention;

[0031] Figure 6 This is a schematic diagram of another detection area image provided in Embodiment 2 of the present invention;

[0032] Figure 7 This is a schematic diagram of a target image provided in Embodiment 2 of the present invention;

[0033] Figure 8 This is a schematic diagram of another target image provided in Embodiment 2 of the present invention;

[0034] Figure 9 This is a schematic diagram of a target image in a multi-target object alternating processing scenario provided in Embodiment 2 of the present invention;

[0035] Figure 10 This is a schematic diagram of a target image in a scenario where multiple target objects are processed simultaneously, as provided in Embodiment 2 of the present invention;

[0036] Figure 11 This is a schematic diagram of the structure of a target object detection device provided in Embodiment 3 of the present invention;

[0037] Figure 12 This is a schematic diagram of another target object detection device provided in Embodiment 4 of the present invention;

[0038] Figure 13 This is a schematic diagram of the structure of a target object detection system provided in Embodiment 5 of the present invention;

[0039] Figure 14 This is a schematic diagram of the structure of an electronic device provided in Embodiment Six of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0041] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices. In the embodiments of this application, certain software, components, models, and other existing industry solutions may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solutions of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0042] The acquisition, transmission, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0043] Example 1

[0044] Figure 1 This is a flowchart of another target object detection method provided in Embodiment 1 of the present invention. This embodiment can be applied to the situation where there is a transmission bandwidth limitation between the acquisition chip and the system chip, and the target object detection method can be executed by a target object detection device. The target object detection device can be implemented in hardware and / or software. The target object detection device can be configured in the system chip and used in conjunction with the acquisition chip.

[0045] like Figure 1 As shown, the method includes:

[0046] S110. Determine at least one current target object region in the current image, and determine the detection region based on the current target object region.

[0047] Here, "current image" refers to the image that the system chip is currently processing, which is acquired and processed by the acquisition chip and then sent to the system chip. "Current target object region" refers to the region where the target object is located in the current image; there can be one or more current target object regions.

[0048] It should be noted that the current image can be either an image directly scaled by the acquisition chip and sent to the system chip, or an image cropped, scaled, and stitched by the acquisition chip based on the historical detection area location and then sent to the system chip.

[0049] Understandably, when a target object first appears in the image captured by the acquisition chip, the chip has no detection area location to refer to. Therefore, it directly scales the captured image and sends it to the system chip. The current image at this point is the image sent to the system chip after the acquisition chip directly scales the captured image. After the system chip detects the target object based on the current image, it uses the technical solution of this embodiment to determine the detection area location and sends the detection area location to the system chip. Therefore, the current image on which the system chip determines the detection area thereafter is the image sent to the system chip after the acquisition chip has cropped, scaled, and stitched the captured image based on historical detection area locations. In this embodiment, the target object can be a face, a human body, a vehicle, a moving object, etc., and this embodiment does not impose any limitations on this. Correspondingly, the detection of the current target object region in the current image can be achieved through a target object detection algorithm or a pre-trained deep learning model.

[0050] The detection region is the area in which the target object may appear in subsequent images, predicted by the system chip based on the current target object region. The detection region can be rectangular in shape, and its size is smaller than the size of the image to be processed.

[0051] In this embodiment, the system chip detects the current target object region in the current image and makes predictions based on the current target object region to obtain the detection region. The purpose of determining the detection region is that the detection region is the area in the image where the target object may appear. The system chip sends the location information of the detection region to the acquisition chip. The acquisition chip performs image cropping, scaling, and stitching based on the detection region and then sends it back to the system chip. At this time, the detection region part of the image still retains the original image quality, and the system chip can perform target object detection on the detection region part, ensuring the target object detection effect.

[0052] Furthermore, in this embodiment, determining the detection area based on the current target object area may include: determining the predicted target object area based on the current target object area; and determining the detection area based on the predicted target object area.

[0053] The predicted target object region can be determined based on factors such as the current target object region, the current target object's movement speed and direction, and the frequency of image acquisition by the acquisition chip. Specifically, the target object region can be predicted using algorithms such as Kalman filtering, optical flow, or pre-trained deep learning models. This embodiment does not limit the specific method used to predict the target object region.

[0054] Understandably, when predicting the target object region, the frequency of image acquisition by the acquisition chip needs to be considered. For example, if the acquisition chip acquires images at a frequency of 10 frames / second, the system chip will predict the location of the current target object region 0.1 seconds later.

[0055] Furthermore, since the acquisition chip needs to process the image after receiving the location of the detection area before sending it to the system chip, the image processing time of the acquisition chip must be considered when predicting the target object region to improve the accuracy of the prediction. The image processing time of the acquisition chip can be added to the frame interval of the image acquisition process to determine the target object region prediction time. For example, if the average image processing time of the acquisition chip is 0.03 seconds, then the prediction of the target object region should be performed on the location of the target object region 0.13 seconds after the current location.

[0056] In an optional embodiment, determining the detection region based on the predicted target object region may include: using the predicted target object region as the detection region.

[0057] In this embodiment, the target object region is predicted based on the current target object region, and the predicted target object region is used as the detection region. This enables the acquisition chip to crop the image after the current image based on the detection region, so that the cropped image contains the target object. At the same time, the cropped image can still maintain the original image quality.

[0058] In another optional embodiment, further determining the detection region based on the predicted target object region may include: taking the union of the current target object region and the predicted target object region as the detection region.

[0059] For example, Figure 2 A schematic diagram of the detection area is provided, such as... Figure 2 As shown, after predicting the target object region based on the current target object region, the current target object's movement speed, and the movement direction, the union of the predicted target object region and the current target object region is used as the detection region.

[0060] In this embodiment, the union of the predicted target object region and the current target object region is used as the detection region. This setting can minimize the missed detection of target objects and ensure the detection effect of target objects.

[0061] Similarly, in this embodiment, when predicting the target object region, it is necessary to consider the frequency of image acquisition by the acquisition chip and the image processing time of the acquisition chip.

[0062] Furthermore, when predicting the target object region, the frequency at which the system chip determines the detection region also needs to be considered. In a specific example, the system chip can determine the detection region frame by frame for each image received from the acquisition chip, once it is determined that a target object region exists. At this point, the target object region is predicted based on the frequency of image acquisition by the acquisition chip and the image processing time of the acquisition chip.

[0063] In another specific example, the system chip can also determine the detection region every preset number of frames or a preset time interval. Since the union of the current target object region and the predicted target object region is used as the detection region in this embodiment, even if the detection region is determined every preset number of frames or a preset time interval, missed detection of the target object can be avoided. In this case, if the system chip determines the detection region every preset number of frames, the target object region is predicted based on the preset number of frames multiplied by the frame interval time, plus the image processing time of the acquisition chip; if the system chip determines the detection region every preset time interval, the target object region is predicted based on the preset time interval and the image processing time of the acquisition chip.

[0064] In another specific example, the system chip can also adjust the frequency of determining the detection area based on the state of the target object. Specifically, if the target object's movement speed is greater than or equal to a preset speed threshold, the detection area is determined frame by frame; if the target object's movement speed is less than the preset speed threshold, the detection area is determined every first preset number of frames or every first preset time interval; if the target object is stationary, the detection area is determined every second preset number of frames or every second preset time interval. The first preset number of frames is less than the second preset number of frames, and the first preset time interval is less than the second preset time interval. This embodiment does not impose limitations on this; in different situations, the methods provided in the above embodiments can be used for corresponding processing.

[0065] Furthermore, determining the detection area based on the current target object area may also include: if the size of the current target object area is determined to be greater than or equal to a preset size threshold, and / or the angle of the target object in the current target object area is within a preset angle range, then the detection area is determined based on the current target object area.

[0066] This embodiment also provides a prerequisite for determining the detection area, namely, determining whether the size and / or angle of the current target object area meet the requirements of target object detection, and then determining the detection area when the requirements of target object detection are met.

[0067] The preset size threshold and preset angle range can be flexibly set according to the needs of the target object detection method, and this embodiment does not impose any restrictions on them.

[0068] Understandably, high-quality target object feature recognition can only be achieved when the target object region is large enough and the target object appears forward or slightly lateral in the image. Only then can a sufficient quantity and quality of target object feature data be extracted, which can then be used for subsequent target object recognition and authentication. Therefore, in this embodiment, the detection region is determined only after the size and / or angle of the current target object region meet the requirements for target object detection, ensuring the effectiveness of target object detection.

[0069] S120. The location of the detection area is sent to the acquisition chip so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image according to the location of the detection area.

[0070] The location of the detection area can be represented by the coordinates of its four vertices, or by the center coordinates and size of the detection area; this embodiment does not impose any restrictions on this. Vertex coordinates or center coordinates can be either absolute coordinates (such as pixel coordinates) or relative coordinates (such as ratios).

[0071] Furthermore, when the detection result of the target object detection algorithm of the system chip is represented in the form of relative coordinates such as ten-thousandths coordinates, the system chip can directly send the detection area position in the form of relative coordinates such as ten-thousandths coordinates to the acquisition chip, and the acquisition chip completes the coordinate transformation of the detection area position into pixel coordinates before performing subsequent processing; alternatively, the system chip can directly convert the detection area position in the form of relative coordinates such as ten-thousandths coordinates into pixel coordinates before sending it to the acquisition chip, and the acquisition chip directly performs subsequent processing based on the detection area position in the form of pixel coordinates. This embodiment does not impose any restrictions on this.

[0072] The image to be processed is at least one image acquired by the acquisition chip after the current image. For example, the acquisition chip sends image P1 to the system chip, where image P1 is the current image. The system chip determines the detection region location based on image P1 and sends it to the acquisition chip. The acquisition chip then crops and processes image P2 based on the received detection region location. The number of images to be processed can be one or more, and the number of images to be processed can correspond to the frequency with which the system chip determines the detection region location. The specific method for determining the number of images to be processed will be described in detail in the following embodiments, and will not be repeated here.

[0073] Cropping refers to cropping the image to be processed based on the location of the detection area. The purpose of cropping is to ensure that the target object is included in the cropped image while maintaining the original image quality. Scaling refers to compressing the length and / or width of the image to be processed. The purpose of scaling is to ensure that the scaled and stitched target image conforms to a preset image size, meeting the bandwidth limitations between the acquisition chip and the system chip. The preset image size is the size of the image that meets the bandwidth limitations between the acquisition chip and the system chip. Stitching refers to combining the cropped image and the scaled image into a single image.

[0074] In this embodiment, the system chip determines the possible location of the target object and sends it to the acquisition chip. When the acquisition chip acquires the image to be processed, it crops the image based on the detected area. The cropped image includes the target object while maintaining the original image quality. Simultaneously, the acquisition chip scales the image to be processed, preserving global information. Finally, the cropped and scaled images are stitched together. On one hand, the cropping, scaling, and stitching of the image based on the detection area by the acquisition chip ensures that the resulting target image meets the bandwidth limitations between the acquisition chip and the system chip. On the other hand, the system chip performs target object detection based on the received target image processed by the acquisition chip, simultaneously grasping global image information and a clear target object region, thus improving the target object detection effect.

[0075] In an optional embodiment, this embodiment provides a target object detection method in a multi-target object scene. Further, S120 may include: if the number of detection regions is determined to be at least two, then selecting a target detection region in each detection region and sending the position of the target detection region to the acquisition chip, so that the acquisition chip performs cropping, scaling, and stitching processing on the image to be processed acquired after the current image based on the position of the target detection region;

[0076] After performing target object detection based on the target image, the process also includes:

[0077] Repeat the process of determining the detection area and selecting the target detection area in each detection area until all detection areas have been processed.

[0078] Repeat the processing operation for each detection area until the condition for stopping alternating processing is met.

[0079] This embodiment provides a method for alternating processing of target objects in a multi-target object scenario. Specifically, target detection regions are selected from each detection region. Selection can be done sequentially based on the region's number, or by sorting and numbering the regions according to the x-coordinate and / or y-coordinate of the center point or a vertex. For example, x-coordinate priority is given, using the coordinates of the lower-left corner vertex of the detection region as a reference. Detection regions with smaller x-coordinates of their lower-left corner vertex are ranked first. If x-coordinates are the same, detection regions with smaller y-coordinates are ranked first. After sorting, each detection region is numbered sequentially.

[0080] The conditions for stopping the alternation can be either that the number of detection areas is at most one, meaning that it is no longer a multi-target object scenario; or that a sufficient number and quality of feature data have been collected for each target object. Specifically, the number of feature data collected for each target object is greater than or equal to a preset quantity threshold, or the quality score of the feature data collected for each target object (the quality score can be determined by factors such as the accuracy, completeness, consistency, and timeliness of the feature data) is greater than or equal to a preset quality score threshold, or the identification or authentication of each target object has been completed based on the collected feature data.

[0081] This embodiment uses a specific example for illustration: The acquisition chip sends image P1 to the system chip. The system chip detects three target objects 1, 2, and 3 in image P1 and determines the detection region positions a, b, and c, respectively. The system chip first sends the detection region position a as the target detection region position. The acquisition chip performs cropping, scaling, and stitching processing on P2 based on the detection region position a, and sends the target image P2' obtained after processing P2 to the system chip. The system chip performs target detection on P2'. If it still detects three target objects 1, 2, and 3, it sends the detection region position b as the target detection region position. The acquisition chip performs cropping, scaling, and stitching processing on P3 based on the detection region position b, and sends the target image P3' obtained after processing P3 to the system chip. The system chip performs target detection on P3'. If three target objects 1, 2, and 3 are still detected, the detection area position c is sent as the target detection area position. The acquisition chip performs cropping, scaling, and stitching processing on P4 based on the detection area position c, and sends the processed target image P4' to the system chip. The above process is repeated until the alternation stops, that is, the number of target objects detected by the system chip is one more than the number of positions, or the identification of target objects 1, 2, and 3 has been completed.

[0082] In this embodiment, an alternating processing method is adopted to process each target object sequentially. The advantage of this setting is that each time, cropping, scaling, and stitching are performed based on only one detection area, which can ensure that the image quality loss of the scaled image to be processed is minimized, thereby enabling the global detection of the image to be taken into account at the same time when detecting the target object.

[0083] In another optional embodiment, this embodiment provides another target object detection method in multi-target object scenarios. Specifically, if the number of detection regions is determined to be at least two, the position of each detection region is sent to the acquisition chip, so that the acquisition chip can perform cropping, scaling, and stitching processing on the image to be processed acquired after the current image according to the position of each detection region.

[0084] This embodiment provides a method for simultaneously processing multiple target objects in a multi-target scenario. Specifically, when the system chip detects multiple target objects, it simultaneously sends the locations of each detection region to the acquisition chip. After receiving the locations of each detection region, the acquisition chip crops the image to be processed according to each detection region location, obtaining multiple cropped images. Then, the original image to be processed is scaled and stitched together with the cropped images to obtain the target image, which is then sent to the system chip. The system chip then performs object recognition and feature data acquisition on the target image.

[0085] Furthermore, when the acquisition chip crops the image to be processed according to the location of each detection area, it only crops out the detection area portion to obtain a new image. Simultaneously, during image stitching, the cropped images from each detection area are placed vertically, or the cropped images can be placed vertically from the bottom to the top of the image according to their length. Specific details are described in the following embodiments, and will not be repeated here.

[0086] The advantage of this setup is that it places the cropped images from each detection region in a single column, thus maximizing the image quality of the scaled portion of the image to be processed. Simultaneously, it allows multiple relatively clear target object regions to coexist in the target image, improving the efficiency of target object detection.

[0087] S130: Receive the target image obtained after the acquisition chip processes the image to be processed, and perform target object detection based on the target image.

[0088] In this embodiment, the acquisition chip performs cropping, scaling, and stitching processing on the image to be processed based on the detection area position sent by the system chip to obtain the target image, and then sends the target image to the system chip. After receiving the target image, the system chip performs target object detection on the target image. The target object detection here uses the same method as the target object detection when determining the current target object region in S110.

[0089] It should be noted that this embodiment uses image transmission between the acquisition chip (AD) and the system chip (SOC) in a DVR device, based on the BT656 transmission protocol, as an example. However, the technical solution of this embodiment can be applied to any image transmission with bandwidth limitations and the need to ensure image quality.

[0090] Figure 3 A schematic diagram of a DVR device is provided, such as... Figure 3 As shown, each acquisition chip (AD) connects to four analog cameras. If the upper limit of the data transmission protocol BT656 between the acquisition chip and the system chip (SOC) is 2 × 1080P × 30, i.e., 2 × 1920 × 1080 × 30 = 124416000, then the transmission amount distributed to each analog camera is 124416000 / 4 = 31104000, approximately 0.5 times that of a 1080 × P × 30 camera or one times that of a 720 × P × 30 camera. If the analog camera is a 1080P × 30 camera, then after the acquisition chip (AD) acquires the image from the analog camera, the target image size transmitted to the SOC needs to be less than or equal to 960 × 1080.

[0091] by Figure 3 For example, in the existing technology, after the acquisition chip AD acquires the image of the analog camera, the image needs to be scaled to 960×1080 to meet the bandwidth limit of the transmission protocol. This results in the overall image clarity of the system chip SOC not being high, which affects the detection effect of the target object.

[0092] Meanwhile, even with existing technologies that reduce the number of analog cameras connected to the AD converter of the acquisition chip, allowing each analog camera to transmit more data, the performance of the system-on-a-chip (SoC) for target object detection cannot be guaranteed due to the continuous updates and iterations in analog camera specifications. For example, if the number of analog cameras connected to each acquisition chip's AD converter is reduced to two, and the analog camera is a 1080P×30 camera, the SoC can receive the original-size image according to the above calculation process; however, if the analog camera is a 5M×20 camera, the image still needs to be scaled before transmission.

[0093] However, the technical solution of this embodiment is not limited by the number of analog cameras or the specifications of the analog cameras, and can fundamentally solve the problem of poor image quality due to transmission bandwidth limitations, which leads to poor target object detection.

[0094] In this embodiment, the system chip determines the detection area and sends it to the acquisition chip. The acquisition chip crops the image to be processed based on the detection area, scales the original image, and then stitches the images together to transmit the resulting target image to the system chip. This target image satisfies the bandwidth limitations, contains the original image of the target object area, and also takes into account global image information, thus improving the detection effect and efficiency of the target object.

[0095] Furthermore, after performing target object detection based on the target image, the method may further include: if it is determined that the score value of the target object feature data in the target image is greater than or equal to a preset score value threshold, then a stop detection region processing instruction is sent to the acquisition chip so that the acquisition chip can scale the image to be processed according to a preset image size.

[0096] The score can be the quantity of feature data of the target object and / or the quality score of the feature data. The specific method for determining the quality score has been described in the above embodiments and will not be repeated here. When the score is determined based on both the quantity and quality score of the target object's feature data, the quantity and quality score can be assigned the same or different weights and then weighted and summed to obtain the score.

[0097] In this embodiment, the system chip performs target object detection on the target image. If the score value of the target object feature data is greater than or equal to a preset score value threshold, it indicates that the feature data of the target object has been collected and can be used for subsequent target object recognition, target object authentication, etc. At this time, the system chip can stop predicting the target object region and determining the detection region, and send a stop detection region processing command to the acquisition chip. After receiving the stop detection region processing command, the acquisition chip stops the operation of cropping the image to be processed according to the detection region, directly scales the image to be processed according to the preset image size, and then sends the scaled image to the system chip.

[0098] The advantage of this setup is that it accelerates the image processing speed of the acquisition chip and improves its image transmission efficiency. At the same time, it minimizes the loss of sharpness in the image being processed.

[0099] The technical solution of this invention involves a system chip (SIC) performing target object recognition on the current image sent by the acquisition chip to obtain the current target object region. Based on this region, a detection region is determined, and its location is sent to the acquisition chip. This allows the acquisition chip to crop, scale, and stitch subsequent images based on the detection region location. The processed target image is then sent back to the SIC, which performs target object detection based on this image. This solves the problem in existing technologies where the acquisition chip scales the image before transmission, resulting in low target object clarity and poor detection performance when the SIC performs target object detection on the received image. The technical solution of this invention ensures the clarity of the target object detection region in the image received by the SIC, even under bandwidth limitations between the acquisition chip and the SIC, thus improving the target object detection effect.

[0100] Example 2

[0101] Figure 4 The flowchart of a target object detection method provided in Embodiment 2 of the present invention is applicable to situations where target object detection is performed under the premise of transmission bandwidth limitation between the acquisition chip and the system chip. The method can be executed by a target object detection device, which can be implemented in hardware and / or software and can be configured in the acquisition chip.

[0102] like Figure 4 As shown, the method includes:

[0103] S210. If the location of the detection area sent by the system chip is determined, the image to be processed is cropped according to the location of the detection area to obtain the detection area image.

[0104] The form of the detection area location and the process by which the system chip determines the location of the detection area have been described in the above embodiments, and will not be repeated here.

[0105] In this embodiment, the system chip determines the location of the detection area and sends it to the acquisition chip. The acquisition chip then crops the image to be processed based on the location of the detection area to obtain the detection area image. The advantage of this setup is that the detection area is a region where target objects may exist. The cropped detection area image retains the original image quality, and its size is smaller than that of the image to be processed. This ensures that target object detection based on the detection area image avoids missed detections and guarantees the detection effect. Simultaneously, it meets the bandwidth limitations between the system chip and the acquisition chip.

[0106] Furthermore, in an optional embodiment, the image to be processed is cropped according to the location of the detection area to obtain a detection area image. Alternatively, the detection area portion of the image to be processed can be directly cropped according to the location of the detection area to obtain the detection area image.

[0107] For example, Figure 5 A schematic diagram of a detection region image is provided, such as... Figure 5 As shown, the size of the image to be processed is 1920×1080. The image to be processed is cropped according to the location of the detection area, and the size of the resulting detection area image is 150×280.

[0108] In another optional embodiment, cropping the image to be sent according to the location of the detection region to obtain a detection region image may further include: cropping a vertical region in the image to be processed that matches the location of the detection region to obtain a detection region image.

[0109] For example, Figure 6 Another schematic diagram of the detection region image is provided, such as... Figure 6 As shown, the size of the image to be processed is 1920×1080. The vertical region where the detection area is located in the image to be processed is cropped, and the size of the resulting detection area image is 150×1080.

[0110] In this embodiment, the advantage of this cropping method is that it ensures the vertical dimension of the cropped detection region image is consistent with the vertical dimension of the image to be processed, thus facilitating subsequent image stitching. However, this embodiment does not limit the cropping method of the detection region image, as long as the detection region image can include the entire detection region in the image to be processed.

[0111] It should be noted that, in this embodiment, when cropping the image to be processed based on the detection area location, the number of images to be processed can correspond to the frequency at which the system chip determines the detection area location. This embodiment does not limit the frequency at which the system chip determines the detection area location or the number of images to be processed by the acquisition chip.

[0112] Specifically, if the system chip determines the detection region location for every frame, then the number of images to be processed is 1. For example, the acquisition chip sends image P1 to the system chip, the system chip determines the detection region location based on image P1 and sends it to the acquisition chip, the acquisition chip crops and processes image P2 based on the received detection region location, the system chip determines the detection region location based on image P2 and sends it to the acquisition chip, the acquisition chip crops and processes image P3 based on the received detection region location, and so on. The advantage of this setup is that it improves the accuracy of the detection region location, thereby improving the detection effect of the target object and avoiding missed detections.

[0113] If the system chip determines the detection region location every preset time interval or every preset number of frames, then the number of images to be processed is the number of images acquired within the preset time interval, or the preset number. For example, the acquisition chip sends image P1 to the system chip, the system chip determines the detection region location based on image P1 and sends it to the acquisition chip, and the system chip determines the detection region location every 5 frames. The acquisition chip performs cropping and subsequent processing on images P2-P5 based on the received detection region location, the system chip determines the detection region location based on image P6 and sends it to the acquisition chip, and the acquisition chip performs cropping and subsequent processing on images P7-P6 based on the received detection region location. 10 The cropping and subsequent processing are then performed, and so on. The advantage of this setup is that it improves the processing speed of the system chip, thereby increasing the detection speed of the target object while ensuring the detection effect.

[0114] If the system chip determines the detection region location for every frame, but decides whether to send the detection region location to the acquisition chip based on whether the target object is stationary or moving (i.e., whether the detection region location has changed), then the images acquired by the acquisition chip before receiving a new detection region location are considered images to be processed. The acquisition chip uses the latest received detection region location to crop and process the images to be processed. For example, the acquisition chip sends image P1 to the system chip. The system chip determines the detection region location based on image P1 and sends it to the acquisition chip. The acquisition chip crops and processes image P2 based on the received detection region location. The system chip determines the detection region location based on image P2. If the detection region location determined by image P2 is the same as that determined by image P1, it does not send the detection region location determined by image P2 to the acquisition chip. When the acquisition chip acquires image P3, it still uses the detection region location corresponding to the received image P1 to crop and process image P3, and so on. The process continues until the acquisition chip receives a new detection area location. By this time, P4 has already been processed, and when P5 is processed, the latest received detection area location is used. The advantage of this setup is that it ensures the detection effect of the target object while improving the processing speed of the system chip.

[0115] S220. The image to be processed is scaled, and the scaled image to be processed is stitched together with the detection area image to obtain the target image.

[0116] Scaling refers to compressing the length and / or width of the image to be processed. The purpose of scaling is to ensure that the target image obtained after scaling and stitching meets the bandwidth limitations between the acquisition chip and the system chip.

[0117] In this embodiment, a detection region image is cropped from the image to be processed, the image to be processed is scaled, and the scaled image to be processed and the detection region image are stitched together to form a target image. This allows the system chip to consider both global image information and image quality of the target object region when performing target object detection on the target image. Simultaneously, the stitched target image meets the bandwidth limitations between the acquisition chip and the system chip.

[0118] Furthermore, scaling the image to be processed may include scaling the image to be processed according to the size of the detection area image and a preset image size.

[0119] The preset image size refers to the size limit of the target image under the bandwidth constraints between the acquisition chip and the system chip. In the example above, the preset image size is 960×1080. Whether the detection area is directly cropped or the corresponding vertical area is cropped, the length of the detection area image is known. The difference between the length in the preset image size and the length of the detection area image is used as the length of the scaled image to be processed; the width in the preset image size is used as the width of the scaled image to be processed.

[0120] by Figure 5 and Figure 6 Taking the detection region image as an example, when the predicted image size is 960×1080, since the length of the detection region image is 150, the image to be processed needs to be scaled to a size of 810×1080.

[0121] In this embodiment, the image to be processed is scaled according to the preset image size and the size of the detection area image, and the scaled image to be processed is stitched together with the detection area image. This allows the size of the stitched target image to match the preset image size, thereby meeting the bandwidth limitation requirements between the acquisition chip and the system chip.

[0122] When stitching the detection region image and the scaled image to be processed, the position of the detection region image within the target image can be preset. Simultaneously, the settings for the detection region image position are pre-synchronized between the acquisition chip and the system chip. This allows the system chip to accurately determine the detection region image and the scaled image to be processed within the target image after receiving it.

[0123] Correspondingly, if adopted Figure 5 The corresponding detection area image cropping method allows setting the detection area image's position in the target image to the bottom right corner. When the height of the detection area image is less than the height of the preset head image size, blank areas can be used to fill in the area above the detection area image, i.e., the top right corner of the target image. Figure 5 Correspondingly, Figure 7 A schematic diagram of a target image is provided, such as... Figure 7 As shown, the target image has a size of 960×1080, the image to be processed has a size of 810×1080, the detection region image has a size of 150×280, the detection region image is located in the lower right position of the target image, and the upper part is a blank area.

[0124] If adopted Figure 6 The corresponding detection region image cropping method allows you to set the position of the detection region image within the target image to the right side of the image. Figure 6 Correspondingly, Figure 8 An illustration of another target image is provided, such as... Figure 8 As shown, the target image has a size of 960×1080, the image to be processed has a size of 810×1080, the detection region image has a size of 150×1080, and the detection region image is located to the right of the target image.

[0125] This embodiment also provides a method for an acquisition chip to determine target images in a multi-target object scenario. Specifically, if the acquisition chip receives only one detection region position, that is, the system chip uses an alternating processing method for each target object, sending the detection region positions sequentially. In this case, for the acquisition chip, as long as only one detection region position is received, regardless of whether there is one or more target objects in the image to be processed, it only needs to perform cropping, scaling, and stitching of the image to be processed based on the detection region positions.

[0126] For example, Figure 9 A schematic diagram is provided for a target image in a scenario where multiple target objects are processed alternately, such as... Figure 9 As shown, the image to be processed contains target objects 1, 2 and 3. The system chip first sends the detection area position corresponding to target object 1 to the acquisition chip. After receiving the detection area position, the acquisition chip crops the vertical area corresponding to the detection area, scales the original image to be processed, and finally stitches them together to obtain the target image.

[0127] If the acquisition chip receives at least two detection area positions, meaning the system chip processes each target object simultaneously, it sends the positions of each detection area to the acquisition chip at the same time. In this case, the acquisition chip only crops the detection area portion, cropping the image to be processed separately for each detection area to obtain individual detection area images. It then determines whether the detection area images, arranged in descending order of length from bottom to top, can be placed in a single column. If so, they are placed in the same column. Otherwise, the detection area images are sorted in ascending order of length. Two lengths are selected that satisfy the following two conditions: the sum of the two lengths is greater than the maximum length, and the sum of the two lengths is the minimum. The detection area images corresponding to these two lengths are placed at the bottom of the image. Then, the remaining detection area images are placed sequentially from bottom to top in descending order of length, thus placing all the detection area images in a single column.

[0128] For example, Figure 10 A schematic diagram is provided for a target image in a scenario where multiple target objects are processed alternately, such as... Figure 10As shown, the system chip sends the detection area positions corresponding to target objects 1, 2, and 3 to the acquisition chip. After receiving the positions of each detection area, the acquisition chip crops the image to be processed to obtain detection area images A, B, and C. The original image to be processed is scaled, and the detection area images A, B, and C are placed in the same column and finally stitched together to obtain the target image.

[0129] S230. The target image is sent to the system chip so that the system chip can perform target object detection based on the target image.

[0130] In this embodiment, the acquisition chip still transmits the target image according to the original transmission protocol. The target image obtained by cropping, scaling, and stitching the image to be processed using the method of this embodiment can meet the transmission bandwidth limitations between the acquisition chip and the system chip.

[0131] In this embodiment, after the system chip receives the target image, it can obtain a detection region image that retains the quality of the original image and a scaled image to be processed that retains the global information of the image. Therefore, by performing target object detection based on the target image, it can extract both global image features and a sufficient number of target object features, thereby improving the detection effect of target object detection.

[0132] The technical solution of this invention involves sending the detection area location to the acquisition chip via a system chip. The acquisition chip then crops the image to be processed based on the detection area location to obtain a detection area image. This image is then scaled, and the scaled image and the detection area image are stitched together to obtain a target image. This target image is then sent to the system chip, which performs target object detection based on it. This solves the problem in existing technologies where the acquisition chip scales the image before transmission, resulting in low clarity of the target object in the image received by the system chip and poor target object detection performance. The technical solution of this invention ensures the clarity of the target object detection area under the bandwidth limitations between the acquisition chip and the system chip, thereby improving the target object detection effect.

[0133] Example 3

[0134] Figure 11 This is a schematic diagram of a target object detection device provided in Embodiment 3 of the present invention. Figure 11 As shown, the device is deployed on a system-on-a-chip and includes:

[0135] The detection region determination module 310 is used to determine at least one current target object region in the current image, and determine the detection region based on the current target object region;

[0136] The detection area location sending module 320 is used to send the location of the detection area to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image according to the location of the detection area.

[0137] The target object detection module 330 is used to receive the target image obtained after the acquisition chip processes the image to be processed, and to perform target object detection based on the target image.

[0138] The technical solution of this invention involves a system chip recognizing target objects in the current image sent by an acquisition chip to obtain the current target object region. Based on this region, a detection area location is determined and sent back to the acquisition chip. This allows the acquisition chip to process subsequent images based on the detection area location. The processed target image is then sent back to the system chip, which performs target object detection based on this image. This solves the problem in existing technologies where the acquisition chip scales the image before transmission, resulting in low target object clarity and poor target object detection performance when the system chip performs target object detection on the received image. The technical solution of this invention ensures the clarity of the target object detection region in the image received by the system chip within the bandwidth limitations between the acquisition chip and the system chip, thereby improving the target object detection effect.

[0139] Based on the above embodiments, optionally, the detection area determination module 310 includes:

[0140] The target object region determination unit is used to determine the target object region based on the current target object region.

[0141] The detection area determination unit is used to determine the detection area based on the predicted target object area.

[0142] Based on the above embodiments, optionally, the detection area determination module 310 includes:

[0143] The detection area determination condition judgment unit is used to determine the detection area based on the current target object area if the size of the current target object area is greater than or equal to a preset size threshold, and / or the angle of the target object in the current target object area is within a preset angle range.

[0144] The device further includes:

[0145] The stop detection region processing condition judgment module is used to send a stop detection region processing command to the acquisition chip if the score value of the feature data of the target object in the target image is greater than or equal to a preset score value threshold, so that the acquisition chip can scale the image to be processed according to the preset image size.

[0146] Based on the above embodiments, optionally, the detection area location sending module 320 includes:

[0147] The target detection region selection unit is used to select a target detection region from each detection region if the number of detection regions is determined to be at least two, and send the position of the target detection region to the acquisition chip so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image according to the position of the target detection region.

[0148] The device further includes:

[0149] The detection area alternation processing module is used to repeatedly execute the operation of determining the detection area and selecting the target detection area in each detection area until the processing of all detection areas is completed.

[0150] The stop alternating processing condition judgment module is used to repeatedly execute the processing operation on each detection area until it is determined that the stop alternating processing condition is met.

[0151] The target object detection device provided in the embodiments of the present invention can execute the target object detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0152] Example 4

[0153] Figure 12 This is a schematic diagram of another target object detection device provided in Embodiment 4 of the present invention. Figure 12 As shown, the device is deployed on the acquisition chip, and the device includes:

[0154] The image to be processed cropping module 410 is used to crop the image to be processed according to the detection area position if the detection area position sent by the system chip is determined, so as to obtain the detection area image.

[0155] The image stitching module 420 is used to scale the image to be processed and stitch the scaled image to be processed with the detection area image to obtain the target image.

[0156] The target image sending module 430 is used to send the target image to the system chip so that the system chip can perform target object detection based on the target image.

[0157] The technical solution of this invention involves sending the detection area location to the acquisition chip via a system chip. The acquisition chip then crops the image to be processed based on the detection area location to obtain a detection area image. This image is then scaled, and the scaled image and the detection area image are stitched together to obtain a target image. This target image is then sent to the system chip, which performs target object detection based on it. This solves the problem in existing technologies where the acquisition chip scales the image before transmission, resulting in low clarity of the target object in the image received by the system chip and poor target object detection performance. The technical solution of this invention ensures the clarity of the target object detection area under the bandwidth limitations between the acquisition chip and the system chip, thereby improving the target object detection effect.

[0158] Based on the above embodiments, optionally, the image cropping module 410 includes:

[0159] The detection region image determination unit is used to crop the vertical region that matches the position of the detection region in the image to be processed to obtain the detection region image;

[0160] Based on the above embodiments, optionally, the image stitching module 420 to be processed includes:

[0161] The scaling processing unit is used to scale the image to be processed according to the size of the detection area image and the preset image size.

[0162] The target object detection device provided in the embodiments of the present invention can execute the target object detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0163] Example 5

[0164] Figure 13 This is a schematic diagram of the structure of a target object detection system provided in Embodiment 5 of the present invention, as shown below. Figure 13 As shown, the target object detection system includes a system chip and a data acquisition chip.

[0165] The system chip is used to determine at least one current target object region in the current image, and based on the current target object region, determine the detection region and send the position of the detection region to the acquisition chip.

[0166] The acquisition chip is used to receive the detection area position sent by the system chip, and to crop the image to be processed according to the detection area position to obtain the detection area image. The image to be processed is at least one image acquired by the acquisition chip after the current image.

[0167] The acquisition chip is used to scale the image to be processed, and then stitch the scaled image to be processed with the detection area image to obtain the target image, and send the target image to the system chip.

[0168] The system chip is used to receive the target image sent by the acquisition chip and perform target object detection based on the target image.

[0169] It should be noted that, Figure 13 This explanation uses the communication transmission between the acquisition chip and the system chip based on the BT656 transmission protocol as an example. However, this target object detection system can be used for any image transmission with bandwidth limitations and the need to ensure image quality.

[0170] This system, through the cooperation between the acquisition chip and the system chip, and under the premise of the transmission bandwidth limitation between the acquisition chip and the system chip, determines the location of the detection area through the system chip, and then performs cropping, scaling and stitching based on the location of the detection area through the acquisition chip, which can ensure the clarity of the target object detection area, thereby improving the detection effect of the target object.

[0171] Example 6

[0172] Figure 14 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0173] like Figure 14As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0174] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0175] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as object detection methods.

[0176] In some embodiments, the target object detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the target object detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the target object detection method by any other suitable means (e.g., by means of firmware).

[0177] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0178] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0179] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0180] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0181] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0182] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0183] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0184] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting a target object, characterized in that, The method is executed by the system chip, and the method includes: Identify at least one current target object region in the current image, and determine the detection region based on the current target object region; The location of the detection area is sent to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image based on the location of the detection area. The receiver chip processes the image to be processed to obtain the target image, and then performs target object detection based on the target image.

2. The method according to claim 1, characterized in that, Based on the current target object area, determine the detection area, including: Based on the current target object region, determine the predicted target object region; The detection area is determined based on the predicted target object area.

3. The method according to claim 1, characterized in that, Based on the current target object area, determine the detection area, including: If it is determined that the size of the current target object region is greater than or equal to the preset size threshold, and / or the angle of the target object in the current target object region is within the preset angle range, then the detection region is determined based on the current target object region; After performing target object detection based on the target image, the process also includes: If the score value of the feature data of the target object in the target image is determined to be greater than or equal to the preset score value threshold, a stop detection region processing command is sent to the acquisition chip so that the acquisition chip can scale the image to be processed according to the preset image size.

4. The method according to any one of claims 1-3, characterized in that, Send the location of the detection area to the acquisition chip, including: If the number of detection areas is determined to be at least two, then the target detection area is selected in each detection area, and the position of the target detection area is sent to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image according to the position of the target detection area. After performing target object detection based on the target image, the process also includes: Repeat the process of determining the detection area and selecting the target detection area in each detection area until all detection areas have been processed. Repeat the processing operation for each detection area until the condition for stopping alternating processing is met.

5. A target object detection method, characterized in that, The method is executed by the acquisition chip, and the method includes: If the location of the detection area sent by the system chip is determined, the image to be processed is cropped according to the location of the detection area to obtain the detection area image; The image to be processed is scaled, and the scaled image to be processed is stitched together with the detection area image to obtain the target image; The target image is sent to the system chip so that the system chip can perform target object detection based on the target image.

6. The method according to claim 5, characterized in that, Based on the location of the detection area, the image to be sent is cropped to obtain the detection area image, including: In the image to be processed, the vertical region that matches the location of the detection region is cropped to obtain the detection region image.

7. The method according to claim 5, characterized in that, Scaling the image to be processed includes: The image to be processed is scaled according to the size of the detection area image and the preset image size.

8. A target object detection device, characterized in that, The device is deployed on a system-on-a-chip, and the device includes: The detection region determination module is used to determine at least one current target object region in the current image, and determine the detection region based on the current target object region; The detection area location sending module is used to send the location of the detection area to the acquisition chip, so that the acquisition chip can perform cropping, scaling and stitching processing on the image to be processed acquired after the current image based on the location of the detection area. The target object detection module is used to receive the target image obtained after the acquisition chip processes the image to be processed, and to perform target object detection based on the target image.

9. A target object detection device, characterized in that, The device is deployed on the acquisition chip, and the device includes: The image to be processed cropping module is used to crop the image to be processed according to the detection area position sent by the system chip if the detection area position is determined to be received, so as to obtain the detection area image. The image stitching module is used to scale the image to be processed and stitch the scaled image to be processed with the detection area image to obtain the target image. The target image sending module is used to send the target image to the system chip so that the system chip can perform target object detection based on the target image.

10. A target object detection system, characterized in that, The system includes a system chip and a data acquisition chip; The system chip is used to determine at least one current target object region in the current image, and determine a detection region based on the current target object region, and send the position of the detection region to the acquisition chip; The acquisition chip is used to receive the detection area position sent by the system chip, and to crop the image to be processed according to the detection area position to obtain the detection area image. The image to be processed is at least one image acquired by the acquisition chip after the current image. The acquisition chip is used to scale the image to be processed, and then stitch the scaled image to be processed with the detection area image to obtain the target image, and send the target image to the system chip. The system chip is used to receive the target image sent by the acquisition chip and perform target object detection based on the target image.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the target object detection method as described in any one of claims 1-7.

12. A storage medium for storing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the target object detection method as described in any one of claims 1-7.

Images