Unloading personnel detection method, device, equipment and storage medium

Abstract

The present invention relates to the field of personnel detection technology, and in particular to a method, device, equipment and storage medium for detecting unloading personnel, the method comprising: obtaining a shooting area image, and extracting features of the shooting area image based on a preset feature identifier; obtaining a feature identifier of an unloading place and a feature identifier of a parking area according to the extraction result, and determining a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a second shooting center point corresponding to the parking area according to the feature identifier of the parking area; determining a reference straight line based on the first shooting center point and the second shooting center point, and constructing an unloading area to be detected based on the reference straight line; upon receiving an unloading personnel detection request, detecting the unloading personnel according to the unloading area to be detected. Compared with the existing method that can only detect the shooting picture, the present invention can determine the unloading area to be detected according to the shooting area image, and detect the unloading area to be detected, thereby improving the detection accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of personnel detection, and in particular to a method, device, equipment and storage medium for detecting unloading personnel. Background Art

[0002] At present, when dangerous goods and valuables are unloaded automatically, relevant staff are generally required to be present to supervise. In order to prevent staff from leaving their posts without authorization during unloading, monitoring robots can be used for detection, that is, monitoring robots can be used to detect whether there are staff in the unloading area.

[0003] However, the existing images captured by the monitoring robot during the actual detection process do not only include the unloading area, which may result in the misjudgment of the presence of staff when there are people in the non-unloading area but not in the unloading area, resulting in low detection accuracy.

[0004] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of the present invention is to provide a method, device, equipment and storage medium for detecting unloading personnel, aiming to solve the technical problem in the prior art that the monitoring robot directly detects the staff on the shooting screen, which leads to the mistaken judgment that the staff is present when there are personnel in the non-unloading area but no staff in the unloading area, thereby resulting in low detection accuracy.

[0006] To achieve the above object, the present invention provides a method for detecting unloading personnel, the method comprising the following steps:

[0007] Acquire a shooting area image, and perform feature extraction on the shooting area image based on a preset feature identifier;

[0008] Obtaining a feature identifier of the unloading place and a feature identifier of the parking area according to the extraction result, and determining a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a second shooting center point corresponding to the parking area according to the feature identifier of the parking area;

[0009] Determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area based on the reference straight line;

[0010] When a request for unloading personnel detection is received, the unloading personnel is detected according to the unloading area to be detected.

[0011] Optionally, the step of determining a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a second shooting center point corresponding to the parking area according to the feature identifier of the parking area includes:

[0012] Determine a first vertex angle scaling invariant feature point corresponding to the unloading place according to the feature identifier of the unloading place, and determine a first shooting center point according to a diagonal line connecting the first vertex angle scaling invariant feature point;

[0013] A second vertex angle scaling invariant feature point corresponding to the parking area is determined based on the parking area feature identifier, and a second shooting center point is determined based on a diagonal line connecting the second vertex angle scaling invariant feature point.

[0014] Optionally, the step of determining the first vertex scaling invariant feature point corresponding to the unloading place according to the unloading place feature identifier comprises:

[0015] Performing mathematical morphological operations according to the characteristic identifiers of the unloading place, and performing contour fitting on the edge contour of the unloading place obtained after the operations;

[0016] Determine the first vertex scaling invariant feature point corresponding to the unloading place based on the fitted boundary contour of the unloading place;

[0017] Accordingly, the step of determining the second vertex scaling invariant feature point corresponding to the parking area based on the parking area feature identifier includes:

[0018] Performing the mathematical morphological operation based on the parking area feature identifier, and performing contour fitting on the parking area edge contour obtained after the operation;

[0019] A second vertex scaling invariant feature point corresponding to the parking area is determined according to the fitted parking area boundary contour.

[0020] Optionally, the step of determining a reference straight line based on the first shooting center point and the second shooting center point, and constructing a to-be-detected unloading area based on the reference straight line includes:

[0021] Determine a reference straight line based on the first shooting center point and the second shooting center point, and determine a reference direction feature vector based on the reference straight line;

[0022] Determine a calibration parameter according to the first vertex angle scaling invariant feature point or the second vertex angle scaling invariant feature point;

[0023] A to-be-detected unloading area is constructed based on the reference direction feature vector and the calibration parameters.

[0024] Optionally, when receiving the unloading personnel detection request, the step of detecting the unloading personnel according to the unloading area to be detected includes:

[0025] Upon receiving a request for unloading personnel detection, obtaining a personnel frame of the unloading personnel according to the unloading area to be detected;

[0026] Determine a compliance area for unloading at the unloading location based on the first shooting center point, and determine a compliance area for parking at the parking area based on the second shooting center point;

[0027] Obtaining a first overlapping area according to the personnel frame and the unloading compliance area, and obtaining a second overlapping area according to the personnel frame and the parking compliance area;

[0028] The unloading personnel is detected based on the first overlapping area and the second overlapping area.

[0029] Optionally, before the step of determining the unloading compliance area of ​​the unloading location based on the first shooting center point and determining the parking compliance area of ​​the parking area based on the second shooting center point, the step further includes:

[0030] Determine the direction information of the unloading place based on the characteristic identification of the unloading place and the characteristic identification of the parking area;

[0031] Determining the unloading port orientation information of the target truck according to the front orientation of the target truck;

[0032] Accordingly, the step of determining the unloading compliance area of ​​the unloading location based on the first shooting center point, and determining the parking compliance area of ​​the parking area based on the second shooting center point includes:

[0033] A compliance unloading area of ​​the unloading area is determined based on the first shooting center point and the unloading area orientation information, and a compliance parking area of ​​the parking area is determined based on the second shooting center point and the unloading port orientation information.

[0034] Optionally, after the step of obtaining a person frame of the unloading personnel according to the unloading area to be detected when receiving the unloading personnel detection request, the method further includes:

[0035] Determining whether a person frame size of the person frame is smaller than a preset person frame size;

[0036] If yes, then obtaining a first distance between the person frame and the first shooting center point, and obtaining a second distance between the person frame and the second shooting center point;

[0037] Determining an occlusion filling direction based on the first distance and the second distance, and filling the person frame based on the occlusion filling direction;

[0038] Accordingly, the step of obtaining a first overlapping area according to the personnel frame and the unloading compliance area, and obtaining a second overlapping area according to the personnel frame and the parking compliance area includes:

[0039] A first overlapping area is obtained based on the filled personnel frame and the unloading compliance area, and a second overlapping area is obtained based on the filled personnel frame and the parking compliance area.

[0040] In addition, to achieve the above-mentioned purpose, the present invention also proposes a device for detecting unloading personnel, the device comprising:

[0041] A feature extraction module, used to obtain a shooting area image and perform feature extraction on the shooting area image based on a preset feature identifier;

[0042] A point determination module, used to obtain a feature identifier of an unloading place and a feature identifier of a parking area according to the extraction result, and determine a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determine a second shooting center point corresponding to the parking area according to the feature identifier of the parking area;

[0043] An area determination module, configured to determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area based on the reference straight line;

[0044] The personnel detection module is used to detect the unloading personnel according to the unloading area to be detected when receiving the unloading personnel detection request.

[0045] In addition, to achieve the above-mentioned purpose, the present invention also proposes a unloading personnel detection device, which includes: a memory, a processor, and an unloading personnel detection program stored on the memory and executable on the processor, and the unloading personnel detection program is configured to implement the steps of the unloading personnel detection method described above.

[0046] In addition, to achieve the above-mentioned purpose, the present invention also proposes a storage medium, on which a unloading personnel detection program is stored, and when the unloading personnel detection program is executed by a processor, the steps of the unloading personnel detection method described above are implemented.

[0047] The present invention first obtains a shooting area image, and performs feature extraction on the shooting area image based on a preset feature identifier; obtains a feature identifier of an unloading place and a feature identifier of a parking area according to the extraction result, and determines a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determines a second shooting center point corresponding to the parking area according to the feature identifier of the parking area; determines a reference straight line based on the first shooting center point and the second shooting center point, and constructs an unloading area to be detected based on the reference straight line; upon receiving a request for unloading personnel detection, detects the unloading personnel according to the unloading area to be detected. Since the present invention can first extract the unloading place feature identifier and the parking area feature identifier in the shooting area image based on the preset feature identifier, and then determine the reference straight line based on the first shooting center point corresponding to the unloading place and the second shooting center point corresponding to the parking area, and construct the unloading area to be detected according to the reference straight line, finally, when the unloading personnel detection request is received, the unloading personnel is detected according to the unloading area to be detected. Compared with the existing method that can only detect the shooting picture, the present invention can determine the unloading area to be detected according to the shooting area image, and detect the unloading area to be detected, thereby improving the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of the structure of a device for detecting unloading personnel in a hardware operating environment according to an embodiment of the present invention;

[0049] Figure 2 This is a flow chart of a first embodiment of a method for detecting unloading personnel according to the present invention;

[0050] Figure 3 This is a schematic diagram of an image of a shooting area in the first embodiment of the method for detecting unloading personnel of the present invention;

[0051] Figure 4 This is a schematic diagram of personnel detection in the first embodiment of the unloading personnel detection method of the present invention;

[0052] Figure 5 This is a flow chart of a second embodiment of the method for detecting unloading personnel of the present invention;

[0053] Figure 6 Schematic diagram of the first shooting center point and the second shooting center point in the second embodiment of the unloading personnel detection method of the present invention;

[0054] Figure 7 It is a flow chart of the third embodiment of the method for detecting unloading personnel of the present invention;

[0055] Figure 8 This is a schematic diagram of an area with good supervisory sight in the third embodiment of the unloading personnel detection method of the present invention;

[0056] Fig. 9 This is a structural block diagram of the first embodiment of the unloading personnel detection device of the present invention.

[0057] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0058] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0059] Reference Figure 1 , Figure 1 The present invention is a schematic diagram of the structure of a device for detecting unloading personnel in a hardware operating environment according to an embodiment of the present invention.

[0060] like Figure 1 As shown, the unloading personnel detection device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk storage. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0061] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the unloading personnel detection device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0062] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a unloading personnel detection program.

[0063] exist Figure 1In the unloading personnel detection device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the unloading personnel detection device of the present invention can be set in the unloading personnel detection device, and the unloading personnel detection device calls the unloading personnel detection program stored in the memory 1005 through the processor 1001, and executes the unloading personnel detection method provided by the embodiment of the present invention.

[0064] It should be noted that, at present, when dangerous goods and valuables are automatically unloaded, relevant staff are generally required to be present to supervise. In order to prevent staff from leaving their posts without authorization during unloading, monitoring robots can be used for detection, that is, monitoring robots can be used to detect whether there are staff in the unloading area.

[0065] However, the existing images captured by the monitoring robot during the actual detection process do not only include the unloading area, which may result in the misjudgment of the presence of staff when there are people in the non-unloading area but not in the unloading area, resulting in low detection accuracy.

[0066] In order to solve the above-mentioned defects, an embodiment of the present invention provides a method for detecting unloading personnel, which can first extract the unloading place feature identifier and the parking area feature identifier in the shooting area image based on the preset feature identifier, and then determine the reference straight line based on the first shooting center point corresponding to the unloading place and the second shooting center point corresponding to the parking area, and construct the unloading area to be detected according to the reference straight line. Finally, when a unloading personnel detection request is received, the unloading personnel is detected according to the unloading area to be detected. Compared with the existing method that can only detect the shooting picture, this embodiment can determine the unloading area to be detected according to the shooting area image, and detect the unloading area to be detected, thereby improving the detection accuracy.

[0067] For ease of understanding, the following combination Figures 2 to 8 The embodiment of the unloading personnel detection method of the present invention and the following embodiments are described.

[0068] refer to Figure 2 , Figure 2 It is a flow chart of the first embodiment of the unloading personnel detection method of the present invention.

[0069] In this embodiment, the unloading personnel detection method includes the following steps:

[0070] Step S10: Acquire a shooting area image, and perform feature extraction on the shooting area image based on a preset feature identifier.

[0071] It should be noted that the method of this embodiment can be applied in the scenario where unloading personnel are detected, or in other scenarios where detection is required. The execution subject of this embodiment can be an unloading personnel detection device with unloading personnel detection, data processing, network communication and program running functions, such as a monitoring robot, or other devices that can achieve the same or similar functions. Here, the above-mentioned unloading personnel detection device (hereinafter referred to as the device) is used to specifically describe this embodiment and the following embodiments.

[0072] It can be understood that in this embodiment, the above-mentioned equipment can be installed in the unloading area of ​​the warehouse. A camera can be set in the above-mentioned equipment, and the unloading area can be photographed by the camera to obtain the above-mentioned photographing area image, wherein the photographing area image may include other areas except the unloading area, and the specific area size is not limited in this embodiment.

[0073] It should be understood that the above-mentioned preset characteristic identifications may include unloading place characteristic identifications and parking area characteristic identifications. In this embodiment, unloading place characteristic identifications are set at the unloading place of the warehouse, and parking area characteristic identifications are set at the parking area where the target truck is parked when unloading. The location of the unloading place can be indicated by the unloading place characteristic identification, and the location of the parking area can be indicated by the parking area characteristic identification. The above-mentioned target truck is the truck that currently needs to be monitored.

[0074] For easier understanding, please refer to Figure 3 To explain, Figure 3 FIG. 1 is a schematic diagram of an image of a shooting area in the first embodiment of the method for detecting unloading personnel of the present invention. Figure 3 As shown, the above device can use the camera to take pictures such as Figure 3 The shooting area image shown in the figure, of course, the shooting angle can be set according to the actual situation, and this embodiment does not limit this. In the shooting area image, there may be a to-be-detected unloading area and a non-unloading area (i.e., an area other than the to-be-detected unloading area), and in the to-be-detected unloading area, there may be an unloading place and a parking area, wherein the unloading place is provided with an unloading place feature mark, and the parking area is provided with a parking area feature mark, such as Figure 3 As shown, the unloading area feature identifier in this embodiment is the identifier corresponding to the four top corners in the top rectangular frame of the unloading area, and the parking area feature identifier is the identifier corresponding to the four top corners in the rectangular frame of the parking area. The above-mentioned unloading area feature identifier and parking area feature identifier can be used as the above-mentioned preset feature identifier.

[0075] In a specific implementation, the above-mentioned device can obtain an image of the shooting area, and extract features from the image of the shooting area based on preset feature identifiers.

[0076] Step S20: Obtain the unloading place feature identifier and the parking area feature identifier according to the extraction result, and determine the first shooting center point corresponding to the unloading place according to the unloading place feature identifier, and determine the second shooting center point corresponding to the parking area according to the parking area feature identifier.

[0077] It should also be noted that the first shooting center point may be the center point corresponding to the rectangular frame of the unloading area, and the second shooting center point may be the center point corresponding to the rectangular frame of the parking area. Since the marks at the four vertices of the rectangle are used as preset feature marks in this embodiment, the intersection of the lines connecting the marks at the diagonal corners of the rectangle can be used as the first shooting center point and the second shooting center point, that is, Figure 3 A is the first shooting center point mentioned above, and B is the second shooting center point mentioned above.

[0078] It is understandable that in order to extract the feature identification of the unloading area and the parking area from the shooting area image, in this embodiment, it can be implemented based on the pedestrian detection convolutional neural network model (such as Posenet series, Yolo series, etc.), that is, several shooting area images with preset feature identifications under different angles, different climatic conditions and other environments are marked as training sets, and then some are selected as test sets to train the pedestrian detection convolutional neural network model, so that it can identify the corresponding area based on the preset feature identification. Of course, other convolutional neural network models can also be selected, and this embodiment is not limited to this.

[0079] It should be emphasized that since the composition of the unloading area characteristic mark and the parking area characteristic mark in this embodiment is consistent, but the size of the unloading area and the parking area is inconsistent, the above-mentioned equipment only needs to identify the area where the preset characteristic mark exists, and then obtain the distance between the preset characteristic marks in each area. The smaller distance can be determined as the unloading area characteristic mark, and the larger distance can be determined as the parking area characteristic mark.

[0080] In a specific implementation, the above-mentioned device can obtain preset feature identifiers in the shooting area image according to the extraction results, and compare the distances between the preset feature identifiers, determine the unloading area feature identifier and the parking area feature identifier according to the comparison results, and determine the first shooting center point according to the connecting line between the unloading area feature identifiers at the diagonals, and determine the second shooting center point according to the connecting line between the parking area feature identifiers at the diagonals.

[0081] Step S30: determining a reference straight line based on the first shooting center point and the second shooting center point, and constructing a to-be-detected unloading area based on the reference straight line.

[0082] It should be understood that the above-mentioned reference straight line can be a connection between the first shooting center point and the second shooting center point. Since the reference is established based on the first shooting center point and the second shooting center point, the center point of the above-mentioned unloading area can be located on the above-mentioned reference straight line. The above-mentioned equipment can construct the unloading area to be detected based on the reference straight line.

[0083] It should be noted that the above unloading area to be inspected may be an area where unloading personnel supervise unloading, such as Figure 3 As shown, the unloading area to be inspected may include a parking area and an unloading place, so that the unloading personnel can effectively supervise the unloading process in the unloading area to be inspected.

[0084] In a specific implementation, the above-mentioned device can determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area including a parking area and an unloading place based on the reference straight line.

[0085] Step S40: upon receiving a request for unloading personnel detection, detecting the unloading personnel according to the unloading area to be detected.

[0086] It is understandable that the above-mentioned unloading personnel detection request can be a request for starting to detect unloading personnel in the unloading area. This embodiment can be triggered by detecting whether the target truck is parked in the parking area, or by detecting whether the unloading area is open. Of course, it can also be other triggering methods, which are not limited by this embodiment.

[0087] Furthermore, considering that the recognition result of the unloading personnel identified by the above device is a three-dimensional rectangular frame, and the result of the unloading area to be detected is a flat rectangular frame, in order to accurately detect the unloading personnel, in this embodiment, the above step S40 includes:

[0088] Step S41: upon receiving a request for unloading personnel detection, obtaining a personnel frame of the unloading personnel according to the unloading area to be detected.

[0089] It should be understood that the above-mentioned personnel frame can be a rectangular frame corresponding to the part where the unloading personnel are in the area to be unloaded. The above-mentioned device can also recognize personnel in the image of the shooting area based on the pedestrian detection convolutional neural network model, and determine the corresponding personnel frame according to the recognition result.

[0090] Step S42: determining a compliance area for unloading at the unloading location based on the first shooting center point, and determining a compliance area for parking at the parking area based on the second shooting center point.

[0091] It should be noted that not all areas in the unloading area to be inspected are areas with good supervision vision. For example, locations far away from the unloading area and the parking area in the unloading area to be inspected may be unfavorable for supervision by the unloading personnel. Therefore, in this embodiment, the unloading compliance area can be determined based on the above-mentioned first shooting center point, and the parking compliance area can be determined based on the second shooting center point.

[0092] It can be understood that in this embodiment, a preset unloading compliance distance can be set based on the first shooting center point, and a preset parking compliance distance can be set based on the second shooting center point, and then a circle is drawn with the first shooting center point as the center and the preset unloading compliance distance as the radius to obtain the unloading compliance area, and a circle is drawn with the second shooting center point as the center and the preset parking compliance distance as the radius to obtain the parking compliance area.

[0093] It should be emphasized that, in this embodiment, a preset unloading compliance area can be set based on the first shooting center point, a preset parking compliance area can be set based on the second shooting center point, and the unloading compliance area can be determined according to the preset unloading compliance area, and the parking compliance area can be determined according to the preset parking compliance area. For ease of understanding, please refer to Figure 4 To explain, Figure 4 FIG. 1 is a schematic diagram of personnel detection in the first embodiment of the unloading personnel detection method of the present invention. Figure 4 As shown, the partial area to the right of the dotted line in the unloading area to be inspected may be the unloading compliance area determined by the first shooting center point according to the preset unloading compliance area, and the partial area to the left of the dotted line may be the parking compliance area determined by the second shooting center point according to the preset parking compliance area. The specific preset unloading compliance area and the preset parking compliance area may be set according to actual conditions.

[0094] Step S43: obtaining a first overlapping area according to the personnel frame and the unloading compliance area, and obtaining a second overlapping area according to the personnel frame and the parking compliance area.

[0095] It should be understood that the first overlapping area may be the area of ​​the overlapping portion between the personnel frame and the unloading compliance area, and the second overlapping area may be the area of ​​the overlapping portion between the personnel frame and the parking compliance area. Figure 4 As shown, Figure 4The figure shows the personnel frames corresponding to two unloading personnel, which are the first personnel frame and the second personnel frame. The head part of the first personnel frame overlaps with the unloading compliance area, and the area corresponding to the head part of the first personnel frame can be used as the first overlapping area; most of the second personnel frame overlaps with the unloading compliance area, and a small part overlaps with the parking compliance area, and the area corresponding to the overlapping part of the second personnel frame and the unloading compliance area can be used as the first overlapping area, and the area corresponding to the overlapping part of the second personnel frame and the parking compliance area can be used as the second overlapping area.

[0096] Step S44: Detecting the unloading personnel based on the first overlapping area and the second overlapping area.

[0097] It should also be noted that this embodiment can judge whether the unloading personnel is in the unloading area to be detected based on the sum of the first overlapping area and the second overlapping area. That is, if the sum of the first overlapping area and the second overlapping area is greater than the preset personnel frame area threshold, it can be determined that the unloading personnel is in the unloading area to be detected. The above-mentioned preset personnel frame area threshold can be set according to actual conditions, and this embodiment does not limit it.

[0098] Continue to combine Figure 4 As shown, it is obvious that the sum of the first overlapping area and the second overlapping area of ​​the first personnel frame may be smaller than the preset personnel frame area threshold, and thus it can be determined that the unloading personnel corresponding to the first personnel frame is not in the unloading area to be detected, and the sum of the first overlapping area and the second overlapping area of ​​the second personnel frame may be larger than the preset personnel frame area threshold, and thus it can be determined that the unloading personnel corresponding to the second personnel frame is in the unloading area to be detected.

[0099] In a specific implementation, when the above-mentioned device receives a request for unloading personnel detection, it can obtain the person frame corresponding to the unloading personnel according to the unloading area to be detected, and determine the unloading compliance area based on the first shooting center point, determine the parking compliance area based on the second shooting center point, and then determine the first overlapping area where the person frame overlaps with the unloading compliance area, and the second overlapping area where the person frame overlaps with the parking compliance area, and detect the unloading personnel based on the sum of the first overlapping area and the second overlapping area, thereby improving the accuracy of the detection.

[0100] In this embodiment, the above-mentioned device can obtain a shooting area image, and perform feature extraction on the shooting area image based on a preset feature identifier; obtain a preset feature identifier in the shooting area image according to the extraction result, compare the distances between the preset feature identifiers, determine the unloading place feature identifier and the parking area feature identifier according to the comparison result, and determine the first shooting center point according to the connecting line between the unloading place feature identifiers at the diagonal positions, and determine the second shooting center point according to the connecting line between the parking area feature identifiers at the diagonal positions; determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area including the parking area and the unloading place based on the reference straight line; upon receiving a detection request for an unloading person, obtain a person frame corresponding to the unloading person according to the to-be-detected unloading area, determine the unloading compliance area based on the first shooting center point, determine the parking compliance area based on the second shooting center point, and then determine a first overlapping area where the person frame overlaps with the unloading compliance area, and a second overlapping area where the person frame overlaps with the parking compliance area, and detect the unloading person based on the sum of the first overlapping area and the second overlapping area. Since the present embodiment can first extract the unloading place feature identifier and the parking area feature identifier in the shooting area image based on the preset feature identifier, and then determine the reference straight line based on the first shooting center point corresponding to the unloading place and the second shooting center point corresponding to the parking area, and construct the unloading area to be detected according to the reference straight line, and finally, when the unloading personnel detection request is received, the unloading personnel is detected according to the unloading area to be detected. Compared with the existing method that can only detect the shooting picture, the present embodiment can determine the unloading area to be detected according to the shooting area image, and only detect in the unloading area to be detected, thereby improving the detection accuracy.

[0101] refer to Figure 5 , Figure 5 It is a flow chart of the second embodiment of the unloading personnel detection method of the present invention.

[0102] In order to accurately identify the first shooting center point and the second shooting center point, Figure 5 As shown, in this embodiment, the above step S20 includes:

[0103] Step S21: determining a first vertex angle scaling invariant feature point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a first shooting center point according to a diagonal line connecting the first vertex angle scaling invariant feature point.

[0104] Step S22: determining a second vertex angle scaling invariant feature point corresponding to the parking area based on the parking area feature identifier, and determining a second shooting center point based on a diagonal line connecting the second vertex angle scaling invariant feature point.

[0105] It should be noted that, in this embodiment, the above-mentioned scale-invariant feature points can use the scale-invariant feature transform (SIFT) algorithm to extract the unloading area feature identifier and the parking area feature identifier from the captured image. In this embodiment, the above-mentioned scale-invariant feature points can be orb feature points, sift feature points, etc. This embodiment uses sift feature points for illustration.

[0106] It can be understood that in order to implement the SIFT algorithm, the above-mentioned device of this embodiment can set a feature point detection module, a feature point extraction module and a feature point description module in the above-mentioned device, wherein the feature point detection module can preprocess the unloading area feature identification and the parking area feature identification, the feature point extraction module can filter out some edges and unstable feature points in the preprocessed unloading area feature identification and the parking area feature identification, the feature point description module can calculate the main direction of each feature point, and construct a feature descriptor based on the field of each feature point, and use the Euclidean distance to calculate the similarity or distance between different feature descriptors, so as to obtain the above-mentioned first vertex angle scaling invariant feature point and the second vertex angle scaling invariant feature point.

[0107] It should be understood that the first vertex scaling invariant feature points may be feature points at the four vertex corners corresponding to the feature mark of the unloading place, and the second vertex scaling invariant feature points may be feature points at the four vertex corners corresponding to the feature mark of the parking area. After determining the first vertex scaling invariant feature points and the second SITF feature points, the centroids of the first vertex scaling invariant feature points and the second vertex scaling invariant feature points may be taken, and the first shooting center point and the second shooting center point may be obtained based on the centroid connection line at the diagonal positions. For ease of understanding, reference may be made to Figure 6 To explain, Figure 6 Schematic diagram of the first shooting center point and the second shooting center point in the second embodiment of the unloading personnel detection method of the present invention, as shown in Figure 6 As shown, the centroids of the first vertex angle scaling invariant feature points can be recorded as M1, M2, M3 and M4 in sequence, and the centroids of the second vertex angle scaling invariant feature points can be recorded as N1, N2, N3 and N4 in sequence. Then, the diagonal lines of the first vertex angle scaling invariant feature points can be the lines between M1M4 and M2M3, and the obtained intersection point A is the first shooting center point. The diagonal lines of the second vertex angle scaling invariant feature points can be the lines between N1N4 and N2N3, and the obtained intersection point B is the second shooting center point.

[0108] Furthermore, in order to accurately detect the first vertex SITF feature point and the second vertex SITF feature point, in this embodiment, the step of determining the first vertex scaling invariant feature point corresponding to the unloading place according to the unloading place feature identifier includes:

[0109] Step S211: performing mathematical morphological operations according to the feature identifiers of the unloading place, and performing contour fitting on the edge contour of the unloading place obtained after the operations;

[0110] Step S212: determining the first vertex scaling invariant feature point corresponding to the unloading place based on the fitted boundary contour of the unloading place;

[0111] Accordingly, the step of determining the second vertex scaling invariant feature point corresponding to the parking area based on the parking area feature identifier includes:

[0112] Step S221: performing the mathematical morphological operation based on the parking area feature identifier, and performing contour fitting on the parking area edge contour obtained after the operation;

[0113] Step S222: determining a second vertex scaling invariant feature point corresponding to the parking area according to the fitted parking area boundary contour.

[0114] It should be noted that, during preprocessing, the above-mentioned device of this embodiment can perform Gaussian convolution processing on the feature identification of the unloading place and the feature identification of the parking area based on the closing operation in mathematical morphology, that is, first dilation and then corrosion. At the same time, in order to implement the Gaussian convolution processing in hardware, this embodiment can select a 5×5 window as the Gaussian convolution for calculation. Of course, other windows can also be selected, and this embodiment does not limit this.

[0115] It is understandable that, considering that the parking area and the unloading area are generally set up outdoors, in order to prevent the influence of light on recognition, in this embodiment, a median filter can be used to cover a 5×5 window on each pixel of the unloading area feature mark and the parking area feature mark, and then the pixels in the window are sorted according to the grayscale value, and the median is taken as the new value of the center pixel, thereby removing some abnormal values ​​generated by the influence of light, making the unloading area feature mark and the parking area feature mark smooth and uniform, thereby improving the accuracy of recognition.

[0116] It should be understood that after the Gaussian convolution calculation is completed, the calculated unloading area feature identifier and parking area feature identifier can be downsampled, and this embodiment can obtain the unloading area feature identifier and parking area feature identifier with different sampling rates by changing the enable end of the data line.

[0117] It should also be noted that since the unloading area feature identifier and the parking area feature identifier in this embodiment are both set as rectangles, the edge contours of the unloading area feature identifier and the parking area feature identifier can be fitted after the above-mentioned mathematical morphology operation, that is, a fixed-size window is set around the feature points obtained after the operation. The specific size of the window is not limited in this embodiment, and the center point of the window is the position of the above-mentioned feature point. The distance between each pixel point and the center point is calculated based on the pixel value at the edge of the window, and the four points with the largest distance are selected. Since the distance between the four vertices in the rectangle is the largest, the four points obtained are the four vertices corresponding to the window, and the edge contour fitting can be completed by connecting the four vertices in sequence.

[0118] It can be understood that after determining the first vertex angle scaling invariant feature point and the second vertex angle scaling invariant feature point, the feature points at two diagonals in the first vertex angle scaling invariant feature point can be connected, and the feature points at two diagonals in the second vertex angle scaling invariant feature point can be connected, and then the intersection of the connections can be the first shooting center point and the second shooting center point mentioned above.

[0119] In a specific implementation, the above-mentioned device can perform Gaussian convolution and down-sampling according to the feature identifier of the unloading place and the feature identifier of the parking area, respectively, to obtain the edge contour of the unloading place and the edge contour of the parking area, respectively, and then perform contour fitting on the edge contour of the unloading place and the edge contour of the parking area to obtain accurate first vertex angle scaling invariant feature points and second vertex angle scaling invariant feature points, and connect the diagonal feature points of the first vertex angle scaling invariant feature points in pairs, and connect the diagonal feature points of the second vertex angle scaling invariant feature points in pairs, to obtain the first shooting center point position and the second shooting center point position.

[0120] Further, in order to accurately determine the unloading area to be detected, in this embodiment, the above step S30 includes:

[0121] Step S31: determining a reference straight line based on the first shooting center point and the second shooting center point, and determining a reference direction feature vector based on the reference straight line.

[0122] It should be understood that the above-mentioned reference direction feature vector may be a vector used to identify the direction of the reference straight line. Figure 6 As shown, after determining the first shooting center point A and the second shooting center point B, the line between A and B can be used as the above-mentioned reference straight line, and the vector corresponding to the direction from A to B can be the above-mentioned reference direction feature vector.

[0123] Step S32: determining calibration parameters according to the first vertex angle scaling invariant feature points or the second vertex angle scaling invariant feature points.

[0124] It should be noted that, in this embodiment, the calibration parameters can be determined based on the coordinates of the first vertex scaling invariant feature point or the coordinates of the second scaling invariant feature point in combination with the world coordinate system, and the calibration parameters can be determined as follows: Figure 6 As shown, for example, when the above-mentioned device detects that the coordinates of N3 are (1, 1), the coordinates of N4 are (6, 1), the coordinates of N1 are (2, 7), and the coordinates of N2 are (4, 7), the world coordinate system is fixed on the actual ground. Then, for the physical coordinates W of any point among N1, N2, N3 and N4 on the actual ground, W=0. Then, calibration can be performed according to the coordinates of each point and the physical coordinates in the world coordinate system to obtain the corresponding intrinsic parameter matrix, extrinsic parameter matrix and distortion parameters. Similarly, the calibration parameters can also be determined according to the coordinates of the second vertex scaling invariant feature point, which will not be repeated in this embodiment.

[0125] Step S33: constructing a to-be-detected unloading area based on the reference direction feature vector and the calibration parameters.

[0126] It can be understood that after determining the calibration parameters and the reference direction feature vector, the world coordinate system of the four vertexes of the unloading area to be detected in the preset world coordinate system can be obtained, and the size and position of the unloading area to be detected in the shooting area image can be obtained based on the world coordinate system of the four vertexes and in combination with the reference direction feature vector and the calibration parameters. That is, in the actual process, the unloading personnel can determine the unloading area to be detected on the actual site in advance according to the actual site conditions, and input the world coordinate system of the four vertexes of the unloading area to be detected into the above-mentioned device, and the above-mentioned device can determine the unloading area to be detected in the shooting area image based on the world coordinate system of the four vertexes and in combination with the reference direction feature vector and the calibration parameters.

[0127] In a specific implementation, the above-mentioned device can determine the reference direction feature vector based on the first shooting center point and the second shooting center point, and determine the calibration parameters based on the first vertex angle scaling invariant feature point or the second vertex angle scaling invariant feature point, and then obtain the world coordinate system at the four vertex corners of the unloading area to be detected in the actual scene, and determine the unloading area to be detected in the shooting area image based on the world coordinate system of the four vertex corners and in combination with the reference direction feature vector and the calibration parameters.

[0128] In this embodiment, the above-mentioned device can perform Gaussian convolution and down sampling according to the feature identifier of the unloading place and the feature identifier of the parking area, respectively, to obtain the edge contour of the unloading place and the edge contour of the parking area, respectively, and then perform contour fitting on the edge contour of the unloading place and the edge contour of the parking area to obtain accurate first vertex scaling invariant feature points and second vertex scaling invariant feature points, and connect the feature points of the first vertex scaling invariant feature points in pairs, and connect the feature points of the second vertex scaling invariant feature points in pairs, to obtain the first shooting center point and the second shooting center point; determine the reference direction feature vector based on the first shooting center point and the second shooting center point, and determine the calibration parameters according to the first vertex scaling invariant feature point or the second vertex scaling invariant feature point, and then obtain the world coordinate system of the four vertex corners of the unloading area to be detected in the actual scene, and determine the unloading area to be detected in the shooting area image based on the world coordinate system of the four vertex corners and in combination with the reference direction feature vector and the calibration parameters.

[0129] refer to Figure 7 , Figure 7 It is a flow chart of the third embodiment of the unloading personnel detection method of the present invention.

[0130] Considering that not all areas within the unloading area to be inspected are areas set up by safety supervision regulations, for example, Figure 8 , Figure 8 Schematic diagram of an area with good supervisory sight in the third embodiment of the unloading personnel detection method of the present invention, as shown in FIG. Figure 8 As shown, the unloading area to be detected on the side of the parking area far from the unloading place may be blocked by the target truck, which is not conducive to monitoring. In order to further improve safety, in this embodiment, before the above step S42, it also includes:

[0131] Step S421: Determine the direction information of the unloading place based on the characteristic identification of the unloading place and the characteristic identification of the parking area.

[0132] It should be noted that the unloading place direction information may be the direction information of receiving goods at the unloading place. In order to facilitate unloading, the direction of receiving goods in the unloading place is generally toward one side of the parking area, that is, Figure 8 The direction from midpoint A to point B.

[0133] It can be understood that the above-mentioned equipment can determine the direction information of the unloading area based on the positional relationship between the unloading area characteristic identifier and the parking area characteristic identifier. If the unloading area characteristic identifier is located on the left side of the parking area characteristic identifier, the direction can be the right direction; if the unloading area characteristic identifier is located on the right side of the parking area characteristic identifier, the direction can be the left direction.

[0134] Step S422: Determine the unloading port orientation information of the target truck according to the front orientation of the target truck.

[0135] It should be understood that due to safety considerations, the unloading port of the target truck is generally set on the right side of the target truck, and then the above-mentioned device can first identify the front direction of the target truck, and determine the direction of the unloading port based on the front direction, such as Figure 8 As shown, if the front of the target truck is facing the direction from point N3 to point N1, the direction of the unloading port may be the direction toward the unloading place.

[0136] Accordingly, the above step S42 includes:

[0137] Step S423: determining a compliance unloading area of ​​the unloading location based on the first shooting center point and the unloading location orientation information, and determining a compliance parking area of ​​the parking area based on the second shooting center point and the unloading port orientation information.

[0138] It should also be noted that the above-mentioned unloading compliance area can be an area that meets the requirements of the unloading place monitored by the unloading personnel, and the above-mentioned parking compliance area can be an area that meets the requirements of the target truck monitored by the unloading personnel.

[0139] It is understandable that after determining the direction information of the unloading place, the above-mentioned device can obtain the preset safety monitoring angle corresponding to the unloading place. The above-mentioned preset safety monitoring angle can be a safety angle detected by the unloading personnel. For example, the preset safety monitoring angle corresponding to the unloading place can be 175°. Figure 8 According to the first shooting center point A, the preset safety monitoring angle corresponding to the unloading place and the direction information of the unloading place, it can be obtained that the 175° fan-shaped area formed between the straight lines K1 and K2 is the above-mentioned unloading compliance area; if the preset safety monitoring angle corresponding to the above-mentioned parking area is 155°, similarly, according to the second shooting center point B, the preset safety monitoring angle corresponding to the unloading place and the direction information of the unloading port, it can be obtained that the 155° fan-shaped area formed between the straight lines K3 and K4 is the above-mentioned parking compliance area. The above-mentioned preset safety monitoring angle can be set according to the actual situation, and this embodiment is not limited.

[0140] It should be understood that after determining the above-mentioned unloading compliance area and parking compliance area, the overlapping part of the unloading compliance area and the parking compliance area can be used as the detection area, that is, Figure 8 The overlapping area between K1, K2, K3 and K4 in the unloading area to be detected, if it is detected that the unloading personnel are in this area, it meets the requirements; if not in this area, it does not meet the requirements.

[0141] In a specific implementation, the above-mentioned equipment can determine the direction information of the unloading place based on the characteristic identification of the unloading place and the characteristic identification of the parking area, determine the direction information of the unloading port based on the direction of the front of the target truck, and then determine the unloading compliance area based on the first shooting center point, the preset safety monitoring angle corresponding to the unloading place, and the direction information of the unloading place, and determine the parking compliance area based on the second shooting center point, the preset safety monitoring angle corresponding to the unloading port, and the direction information of the unloading port, thereby improving safety.

[0142] Furthermore, considering that when the unloading personnel are in the overlapping area between K1, K2, K3 and K4, but are blocked by other objects (such as goods, etc.), resulting in the personnel frame being smaller than the preset personnel frame area threshold, a false alarm occurs, therefore, in this embodiment, after the above step S41, the following is further included:

[0143] Step S411: determining whether the size of the personnel frame is smaller than a preset personnel frame size;

[0144] Step S412: If yes, a first distance between the person frame and the first shooting center point is obtained, and a second distance between the person frame and the second shooting center point is obtained.

[0145] It should be noted that if the body of the unloading personnel is partially blocked by other objects, the size of the obtained personnel frame may be smaller. The above-mentioned device can judge whether obstruction occurs by comparing it with the preset personnel frame size. The specific preset personnel frame size can be set according to actual conditions, and this embodiment does not limit it.

[0146] It is understandable that the first distance may be the distance between the first shooting center point and the center point of the personnel frame, the second distance may be the distance between the second shooting center point and the center point of the personnel frame, and the center point of the personnel frame may be the point corresponding to the center of the personnel frame.

[0147] Step S413: determining an occlusion filling direction based on the first distance and the second distance, and filling the person frame based on the occlusion filling direction.

[0148] It should be understood that after determining the first distance and the second distance, the current position of the person frame can be determined, for example, Figure 8 As shown, if the personnel frame is located above between M1 and M2, is blocked by the goods at the unloading area and the corresponding size is lower than the preset personnel frame size, the above-mentioned first distance and second distance can be obtained. According to the first distance and the second distance, it can be determined that the possible obstructions near the personnel frame are the goods in the unloading area, and then it can be determined that the personnel frame may be the lower body of the unloading personnel blocked by the unloading goods. Therefore, the filling direction can be the direction of filling the personnel frame of the unloading personnel downwards, and then the personnel frame can be filled based on the filling direction.

[0149] Accordingly, the above step S43 includes:

[0150] Step S414: obtaining a first overlapping area based on the filled personnel frame and the unloading compliance area, and obtaining a second overlapping area based on the filled personnel frame and the parking compliance area.

[0151] It should also be noted that, since the above-mentioned personnel frame is filled, the first overlapping area and the second overlapping area obtained will not be affected by the obstruction and result in a smaller area, thereby avoiding false alarms caused by obstruction and improving accuracy.

[0152] In a specific implementation, the above-mentioned device can determine whether the size of the personnel frame is lower than the preset personnel frame size. If so, the first distance and the second distance can be determined based on the personnel frame, the first shooting center point and the second shooting center point, and the positional relationship between the obstruction and the personnel frame can be determined based on the first distance and the second distance. The occlusion filling direction is determined according to the positional relationship, and the personnel frame is filled based on the occlusion filling direction. Then, the first overlapping area and the second overlapping area can be determined based on the filled personnel frame, thereby improving accuracy.

[0153] In this embodiment, the above-mentioned device can determine the direction information of the unloading place based on the characteristic identification of the unloading place and the characteristic identification of the parking area, determine the direction information of the unloading port based on the direction of the front of the target truck, and then determine the unloading compliance area based on the first shooting center point, the preset safety monitoring angle corresponding to the unloading place and the direction information of the unloading place, and determine the parking compliance area based on the second shooting center point, the preset safety monitoring angle corresponding to the unloading port and the direction information of the unloading port, thereby improving safety; at the same time, the above-mentioned device can determine whether the personnel frame size of the personnel frame is lower than the preset personnel frame size. If so, the first distance and the second distance can be determined based on the personnel frame, the first shooting center point and the second shooting center point, the positional relationship between the obstruction and the personnel frame can be determined based on the first distance and the second distance, the obstruction filling direction can be determined according to the positional relationship, the personnel frame can be filled based on the obstruction filling direction, and then the first overlapping area and the second overlapping area can be determined according to the filled personnel frame, thereby improving accuracy.

[0154] In addition, an embodiment of the present invention further proposes a storage medium, on which a unloading personnel detection program is stored. When the unloading personnel detection program is executed by a processor, the steps of the unloading personnel detection method described above are implemented.

[0155] In addition, refer to Fig. 9 , Fig. 9 This is a structural block diagram of the first embodiment of the unloading personnel detection device of the present invention. The embodiment of the present invention further provides an unloading personnel detection device, and the unloading personnel detection device includes:

[0156] The feature extraction module 901 is used to obtain a shooting area image and perform feature extraction on the shooting area image based on a preset feature identifier;

[0157] The point determination module 902 is used to obtain the feature identifier of the unloading place and the feature identifier of the parking area according to the extraction result, and determine the first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determine the second shooting center point corresponding to the parking area according to the feature identifier of the parking area;

[0158] An area determination module 903 is used to determine a reference straight line based on the first shooting center point and the second shooting center point, and to construct a to-be-detected unloading area based on the reference straight line;

[0159] The personnel detection module 904 is used to detect the unloading personnel according to the unloading area to be detected when receiving the unloading personnel detection request.

[0160] In this embodiment, the above-mentioned device can obtain a shooting area image, and perform feature extraction on the shooting area image based on a preset feature identifier; obtain a preset feature identifier in the shooting area image according to the extraction result, compare the distances between the preset feature identifiers, determine the unloading place feature identifier and the parking area feature identifier according to the comparison result, and determine the first shooting center point according to the connecting line between the unloading place feature identifiers at the diagonal positions, and determine the second shooting center point according to the connecting line between the parking area feature identifiers at the diagonal positions; determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area including the parking area and the unloading place based on the reference straight line; upon receiving a detection request for an unloading person, obtain a person frame corresponding to the unloading person according to the to-be-detected unloading area, determine the unloading compliance area based on the first shooting center point, determine the parking compliance area based on the second shooting center point, and then determine a first overlapping area where the person frame overlaps with the unloading compliance area, and a second overlapping area where the person frame overlaps with the parking compliance area, and detect the unloading person based on the sum of the first overlapping area and the second overlapping area. Since the present embodiment can first extract the unloading place feature identifier and the parking area feature identifier in the shooting area image based on the preset feature identifier, and then determine the reference straight line based on the first shooting center point corresponding to the unloading place and the second shooting center point corresponding to the parking area, and construct the unloading area to be detected according to the reference straight line, and finally, when the unloading personnel detection request is received, the unloading personnel is detected according to the unloading area to be detected. Compared with the existing method that can only detect the shooting area, the present embodiment can determine the unloading area to be detected according to the shooting area image, and only detect in the unloading area to be detected, thereby improving the detection accuracy.

[0161] Other embodiments or specific implementations of the unloading personnel detection device of the present invention can refer to the above-mentioned method embodiments and will not be described in detail here.

[0162] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0163] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. Through the description of the above implementation modes, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0164] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for detecting unloading personnel, characterized in that: The method comprises the following steps: Acquire a shooting area image, and perform feature extraction on the shooting area image based on a preset feature identifier; Obtaining a feature identifier of the unloading place and a feature identifier of the parking area according to the extraction result, and determining a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a second shooting center point corresponding to the parking area according to the feature identifier of the parking area; Determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area based on the reference straight line; Upon receiving a request for unloading personnel detection, detecting the unloading personnel according to the unloading area to be detected; The step of determining a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determining a second shooting center point corresponding to the parking area according to the feature identifier of the parking area comprises: Determine a first vertex angle scaling invariant feature point corresponding to the unloading place according to the feature identifier of the unloading place, and determine a first shooting center point according to a diagonal line connecting the first vertex angle scaling invariant feature point; A second vertex angle scaling invariant feature point corresponding to the parking area is determined based on the parking area feature identifier, and a second shooting center point is determined based on a diagonal line connecting the second vertex angle scaling invariant feature point.

2. The unloading personnel detection method according to claim 1, characterized in that: The step of determining the first vertex scaling invariant feature point corresponding to the unloading place according to the unloading place feature identifier comprises: Performing mathematical morphological operations according to the characteristic identifiers of the unloading place, and performing contour fitting on the edge contour of the unloading place obtained after the operations; Determine the first vertex scaling invariant feature point corresponding to the unloading place based on the fitted boundary contour of the unloading place; Accordingly, the step of determining the second vertex scaling invariant feature point corresponding to the parking area based on the parking area feature identifier includes: Performing the mathematical morphological operation based on the parking area feature identifier, and performing contour fitting on the parking area edge contour obtained after the operation; A second vertex scaling invariant feature point corresponding to the parking area is determined according to the fitted parking area boundary contour.

3. The unloading personnel detection method according to claim 1 or 2, characterized in that: The step of determining a reference straight line based on the first shooting center point and the second shooting center point, and constructing a to-be-detected unloading area based on the reference straight line includes: Determine a reference straight line based on the first shooting center point and the second shooting center point, and determine a reference direction feature vector based on the reference straight line; Determine a calibration parameter according to the first vertex angle scaling invariant feature point or the second vertex angle scaling invariant feature point; A to-be-detected unloading area is constructed based on the reference direction feature vector and the calibration parameters.

4. The unloading personnel detection method according to claim 1, characterized in that: The step of detecting the unloading personnel according to the unloading area to be detected upon receiving the unloading personnel detection request comprises: Upon receiving a request for unloading personnel detection, obtaining a personnel frame of the unloading personnel according to the unloading area to be detected; Determine a compliance area for unloading at the unloading location based on the first shooting center point, and determine a compliance area for parking at the parking area based on the second shooting center point; Obtaining a first overlapping area according to the personnel frame and the unloading compliance area, and obtaining a second overlapping area according to the personnel frame and the parking compliance area; The unloading personnel is detected based on the first overlapping area and the second overlapping area.

5. The method for detecting unloading personnel according to claim 4, characterized in that: Before the step of determining the unloading compliance area of ​​the unloading location based on the first shooting center point and determining the parking compliance area of ​​the parking area based on the second shooting center point, the step further includes: Determine the direction information of the unloading place based on the characteristic identification of the unloading place and the characteristic identification of the parking area; Determining the unloading port orientation information of the target truck according to the front orientation of the target truck; Accordingly, the step of determining the unloading compliance area of ​​the unloading location based on the first shooting center point, and determining the parking compliance area of ​​the parking area based on the second shooting center point includes: The unloading compliance area of ​​the unloading place is determined based on the first shooting center point and the unloading place direction information, and the parking compliance area of ​​the parking area is determined based on the second shooting center point and the unloading port direction information.

6. The method for detecting unloading personnel according to claim 4, characterized in that: After the step of obtaining a person frame of the unloading personnel according to the unloading area to be detected upon receiving the unloading personnel detection request, the method further includes: Determining whether a person frame size of the person frame is smaller than a preset person frame size; If yes, then obtaining a first distance between the person frame and the first shooting center point, and obtaining a second distance between the person frame and the second shooting center point; Determining an occlusion filling direction based on the first distance and the second distance, and filling the person frame based on the occlusion filling direction; Accordingly, the step of obtaining a first overlapping area according to the personnel frame and the unloading compliance area, and obtaining a second overlapping area according to the personnel frame and the parking compliance area includes: A first overlapping area is obtained based on the filled personnel frame and the unloading compliance area, and a second overlapping area is obtained based on the filled personnel frame and the parking compliance area.

7. A device for detecting unloading personnel, characterized in that: The device comprises: A feature extraction module, used to obtain a shooting area image and perform feature extraction on the shooting area image based on a preset feature identifier; A point determination module, used to obtain a feature identifier of an unloading place and a feature identifier of a parking area according to the extraction result, and determine a first shooting center point corresponding to the unloading place according to the feature identifier of the unloading place, and determine a second shooting center point corresponding to the parking area according to the feature identifier of the parking area; An area determination module, configured to determine a reference straight line based on the first shooting center point and the second shooting center point, and construct a to-be-detected unloading area based on the reference straight line; A personnel detection module, configured to detect the unloading personnel according to the unloading area to be detected upon receiving a detection request of the unloading personnel; The point determination module is also used to determine the first vertex angle scaling invariant feature point corresponding to the unloading place according to the unloading place feature identifier, and determine the first shooting center point according to the diagonal connection line of the first vertex angle scaling invariant feature point; determine the second vertex angle scaling invariant feature point corresponding to the parking area based on the parking area feature identifier, and determine the second shooting center point based on the diagonal connection line of the second vertex angle scaling invariant feature point.

8. A device for detecting unloading personnel, characterized in that: The device comprises: a memory, a processor and a unloading personnel detection program stored in the memory and executable on the processor, wherein the unloading personnel detection program is configured to implement the steps of the unloading personnel detection method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium stores a unloading personnel detection program, and when the unloading personnel detection program is executed by the processor, the steps of the unloading personnel detection method according to any one of claims 1 to 6 are implemented.

Images