Road construction anti-collision early warning method based on visual identification

Abstract

The invention relates to the technical field of construction safety, and discloses a road construction anti-collision early warning method based on visual identification, which comprises the following steps: when a camera shoots a second device or a constructor, identifying a relative position between the second device or the constructor and a current first device according to a current camera image; the relative position comprises a transverse distance and a longitudinal distance between the second equipment or the constructor and the current first equipment; obtaining current route information of the first equipment and the second equipment or constructors in the sensor; according to the route information, estimating a current estimated route of the first equipment and the second equipment or the constructors; judging whether the pre-estimated route of the second equipment or the constructor coincides with the dangerous area on the pre-estimated route of the current first equipment or not; and under the condition of coincidence, giving an alarm to the current first equipment and second equipment or constructors at the same time. And collision between the construction equipment and constructors can be prevented, and collision between the construction equipment can be avoided, so that the possibility of safety accidents on a construction site is reduced.

Description

Technical Field

[0001] The present application relates to the field of construction safety technology, for example, to a road construction anti-collision warning method based on visual recognition. Background Art

[0002] During road construction, there are situations where construction workers and construction machinery are simultaneously working at the same construction site. Construction machinery, such as rollers, is huge, and drivers have blind spots. It is also difficult for construction workers to always pay attention to the movements of construction machinery when working. If the driver and construction workers are not careful, there is a risk of collision between man and machine. There is also a risk of collision between the two construction machines due to overlapping routes. Especially at night, in tunnels and other places with poor lighting conditions, the risk of collision between man and machine is greater, causing inevitable personal safety, property losses and accidents. Therefore, a road construction anti-collision warning method based on visual recognition is needed to solve the above problems.

[0003] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present application, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0004] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical components or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0005] The embodiment of the present disclosure provides a road construction anti-collision warning method based on visual recognition, including: for all running equipment and construction personnel at the construction site, the equipment is equipped with sensors and cameras are arranged around the equipment; the construction personnel are equipped with sensors; the method includes:

[0006] When the camera captures the second device or the construction worker, the relative position between the second device or the construction worker and the current first device is identified according to the current camera image, wherein the relative position includes a lateral distance and a longitudinal distance between the second device or the construction worker and the current first device;

[0007] Obtaining route information of the current first device and the second device or the construction personnel in the sensor;

[0008] estimating the estimated routes of the current first device and the second device or the construction personnel according to the route information;

[0009] Determine whether the estimated route of the second device or the construction personnel overlaps with the dangerous area on the estimated route of the current first device;

[0010] In the event of overlap, an alarm is simultaneously issued to the current first device and the second device or the construction personnel.

[0011] In one embodiment, the running device is provided with a reference object with scale and direction and can be photographed by a camera. After the camera is installed, the camera is calibrated to obtain the camera internal parameters; when the camera photographs the second device or the construction personnel, the relative position between the second device or the construction personnel and the current first device is identified according to the camera image, including:

[0012] The distortion of the reference object image captured is corrected according to the camera internal parameters and the reference object, and the corresponding relationship between the pixel coordinates of the corrected image and the scale and direction of the reference object is generated;

[0013] Correct the current camera image according to the correspondence between the camera internal parameters and the pixel coordinates of the calibrated image and the scale and direction of the reference object;

[0014] Acquire a target image of the corrected current camera image, where the target image includes a second device or a construction worker;

[0015] The relative position between the target image and the current first device is calculated according to the target image and the reference object.

[0016] In one embodiment, calculating the relative position between the target image and the current first device according to the target image and the reference object includes:

[0017] Segment the target image and the camera image according to a preset segmentation model, and generate the edge of the target image;

[0018] Identify the left and right endpoint pixels of the edge of the target image and record them;

[0019] Determine the endpoint pixel closest to the current first device among the left and right endpoint pixels according to the reference object;

[0020] The horizontal distance and the vertical distance between the nearest endpoint pixel and the current first device are calculated according to the reference object.

[0021] In one embodiment, estimating the estimated routes of the current first device and the second device or the construction personnel according to the route information includes:

[0022] Identify the current working route distribution and cycle of the first device and the second device or the construction personnel and the corresponding time according to the route information;

[0023] Evaluate the current working scope of the first equipment and the second equipment or the construction personnel according to the distribution of the working routes;

[0024] Recording the approximate overlapping area of ​​the working range of the current first equipment and the second equipment or the construction personnel within the corresponding time period, wherein the approximate overlapping area includes a completely overlapping area and a preset approximate danger area;

[0025] The period information of the approximately overlapping area is recorded, and the routes and arrival times of the current first device and the second device or the construction personnel are estimated according to the period information of the approximately overlapping area.

[0026] In one embodiment, estimating the estimated route of the current first device and the second device or the construction personnel according to the route information further includes:

[0027] Acquire multiple frames of current camera images continuously without approximately overlapping areas;

[0028] Segmenting a target image in each frame of the current camera image according to a preset segmentation algorithm, wherein the target image includes the second device or the construction personnel;

[0029] Obtain image information of a set part in a target image;

[0030] The estimated route of the target image is determined based on the image information of the set part of the target image in multiple consecutive frames of current camera images.

[0031] In one embodiment, when the target image is a construction worker, the set parts include feet, legs and upper body; the step of determining the estimated route of the target image according to the image information of the set parts of the target image in the continuous multiple frames of current camera images includes:

[0032] Compare the set part with the preset walking image to identify the moving direction of the target image;

[0033] Obtaining the relative position between the set part and the current first device;

[0034] Calculating the moving speed according to the change in relative position between the set part and the current first device in multiple consecutive frames of current camera images;

[0035] Evaluate the estimated routes of construction crews based on direction of travel and speed of movement.

[0036] In one embodiment, when the target image is a second device, the set part includes the endpoint pixel of the target image that is closest to the current first device; the estimated route of the target image is determined based on the image information of the set part of the target image in multiple consecutive frames of current camera images, including: evaluating the travel direction and moving speed of the second device based on the change in the relative position between the set part and the current first device in multiple consecutive frames of current camera images.

[0037] In one embodiment, the danger zone includes an area where the current first device is located, a first danger zone, and a second danger zone, wherein the first danger zone is close to the area where the current first device is located, and the second danger zone is outside the first danger zone; in the case of overlap, simultaneously issuing an alarm to the current first device and the second device or the construction personnel includes:

[0038] In case of overlap with the second danger zone, the alarm will sound a level-shifted alarm;

[0039] In the case of overlap with the first danger zone, the alarm emits a modified tone alarm and a voice alarm to remind the location of the target that is about to collide.

[0040] The road construction anti-collision warning method based on visual recognition provided by the embodiment of the present disclosure can achieve the following technical effects:

[0041] When the camera captures the second device or construction personnel, it indicates that the second device or construction personnel is close to the current first device. At this time, after evaluating the routes of the current first device, the second device or construction personnel, if there is overlap in the dangerous area, an alarm will be issued to the current first device and the beneficial device or construction personnel. This helps prevent collisions between construction equipment and construction personnel, and also helps avoid collisions between construction equipment, thereby reducing the possibility of safety accidents at the construction site, improving the safety of the construction site, and avoiding personal safety, property losses and accidents.

[0042] The above general description and the following description are exemplary and explanatory only and are not intended to limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] One or more embodiments are exemplarily described by corresponding drawings, which do not limit the embodiments. Elements with the same reference numerals in the drawings are shown as similar elements, and the drawings do not constitute a scale limitation, and wherein:

[0044] Figure 1 It is a schematic diagram of a road construction anti-collision warning method based on visual recognition provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0045] In order to be able to understand the features and technical contents of the embodiments of the present disclosure in more detail, the implementation of the embodiments of the present disclosure is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the embodiments of the present disclosure. In the following technical description, for the convenience of explanation, a full understanding of the disclosed embodiments is provided through multiple details. However, one or more embodiments can still be implemented without these details. In other cases, to simplify the drawings, well-known structures and devices can be simplified for display.

[0046] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0047] Unless otherwise stated, the term "plurality" means two or more.

[0048] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0049] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0050] The term "correspondence" may refer to an association relationship or a binding relationship. The correspondence between A and B means that there is an association relationship or a binding relationship between A and B.

[0051] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.

[0052] In order to collect relevant data, all operating equipment and construction personnel at the construction site are equipped with sensors and cameras around the equipment; the construction personnel are equipped with sensors. The sensors here include sensors that detect moving positions.

[0053] Combination Figure 1 As shown, based on the above-mentioned collected relevant data, the road construction anti-collision warning method based on visual recognition provided by the embodiment of the present disclosure includes:

[0054] S101, when the camera captures the second device or the construction worker, the relative position between the second device or the construction worker and the current first device is identified according to the current camera image, where the relative position includes the lateral distance and the longitudinal distance between the second device or the construction worker and the current first device.

[0055] Here, the horizontal direction refers to the width direction of the current first device, and the vertical direction refers to the length direction of the current first device. The horizontal direction and the vertical direction are perpendicular in a plane.

[0056] S102, obtaining route information of the current first device and the second device or the construction personnel in the sensor.

[0057] S103, estimating the estimated routes of the current first device and the second device or the construction personnel according to the route information.

[0058] S104, determining whether the estimated route of the second device or the construction worker overlaps with the dangerous area on the estimated route of the current first device.

[0059] S105, when there is overlap, an alarm is simultaneously issued to the current first device and the second device or the construction personnel.

[0060] When the camera captures the second device or construction personnel, it indicates that the second device or construction personnel is close to the current first device. At this time, after evaluating the routes of the current first device, the second device or construction personnel, if there is overlap in the dangerous area, an alarm will be issued to the current first device and the beneficial device or construction personnel. This helps prevent collisions between construction equipment and construction personnel, and also helps avoid collisions between construction equipment, thereby reducing the possibility of safety accidents at the construction site, improving the safety of the construction site, and avoiding personal safety, property losses and accidents.

[0061] Optionally, taking into account that the image may be distorted when the camera is shooting, a reference object with a scale and direction is provided on the running device and can be photographed by the camera. After the camera is installed, the camera is calibrated to obtain the camera intrinsic parameters; when the camera photographs the second device or the construction worker, the relative position between the second device or the construction worker and the current first device is identified according to the camera image, including: correcting the distortion of the captured reference object image according to the camera intrinsic parameters and the reference object, and generating a correspondence between the pixel coordinates of the corrected image and the scale and direction of the reference object; correcting the current camera image according to the camera intrinsic parameters and the correspondence between the pixel coordinates of the corrected image and the scale and direction of the reference object; obtaining a target image of the corrected current camera image, the target image including the second device or the construction worker; and calculating the relative position between the target image and the current first device according to the target image and the reference object.

[0062] After calibrating the camera's internal parameters, the distortion of the reference image is corrected by the scale and direction of the reference object, and then the distortion of the entire image is corrected according to the relevant parameters, so that the shape of the actual target image can be restored, which facilitates the subsequent accurate calculation of the lateral and longitudinal distances in the relative position, and helps to improve the accuracy of the alarm.

[0063] Optionally, the relative position between the target image and the current first device is calculated based on the target image and the reference object, including: segmenting the target image and the camera image according to a preset segmentation model to generate the edge of the target image; identifying and recording the left and right endpoint pixels of the edge of the target image; judging the endpoint pixel closest to the current first device among the left and right endpoint pixels based on the reference object; and calculating the horizontal distance and vertical distance between the closest endpoint pixel and the current first device based on the reference object.

[0064] The segmentation model used in this embodiment can be a computer vision semantic segmentation model. Among them, the computer vision semantic segmentation model can be trained by the following method: collect a training image dataset X of the current camera image; use the LabelMe annotation platform to annotate the edge label Y of the training image X; use the FCN (full convolution deep semantic segmentation model) pre-trained on the PASCAL VOC dataset containing large-scale semantic segmentation images of human bodies and mechanical equipment, with the training image dataset X of the current camera image as input, and the edge label Y of human bodies and mechanical equipment as the prediction target; construct a loss function for the computer vision semantic segmentation model; based on stochastic gradient descent, iteratively update the semantic segmentation model parameters end-to-end to obtain a trained computer vision semantic segmentation model.

[0065] After segmenting the target image and the camera image through a computer vision semantic segmentation model, the endpoint pixel point closest to the current first device, that is, the point where a collision may occur, is found. The lateral and longitudinal distances between the point where a collision may occur and the current first device are calculated as relevant parameters of the relative position, which helps to accurately estimate the overlapping areas on the route, thereby improving the accuracy of the alarm.

[0066] Optionally, the estimated routes of the current first device and the second device or the construction personnel are estimated based on the route information, including: identifying the work route distribution and cycle and the corresponding time of the current first device and the second device or the construction personnel based on the route information; evaluating the work scope of the current first device and the second device or the construction personnel based on the work route distribution; recording the approximate overlapping area of ​​the work scope of the current first device and the second device or the construction personnel within the corresponding time period, the approximate overlapping area including the completely overlapping area and the preset approximate danger zone; recording the cycle information of the approximate overlapping area, and estimating the route and arrival time of the current first device and the second device or the construction personnel based on the cycle information of the approximate overlapping area.

[0067] Considering that the working equipment and construction personnel at a general construction site have their own specific construction areas, the working range and cycle of the current first equipment and second equipment or construction personnel are identified and the relevant time is recorded. The overlapping parts in the corresponding time period are regarded as the area where collision may occur, that is, the approximate overlapping area, and the routes and arrival times of the equipment and personnel in the approximate overlapping area are mainly evaluated. In this way, the evaluation scope is narrowed to the approximate overlapping area, the amount of data processed is reduced, and the speed of evaluation is improved.

[0068] Optionally, estimating the estimated routes of the current first device and the second device or the construction personnel based on the route information also includes: acquiring multiple consecutive frames of current camera images without an approximately overlapping area; segmenting a target image in each frame of the current camera image according to a preset segmentation algorithm, the target image including the second device or the construction personnel; acquiring image information of a set portion in the target image; and determining the estimated route of the target image based on the image information of the set portion of the target image in multiple consecutive frames of the current camera image.

[0069] Taking into account special situations such as the possible displacement of equipment or construction personnel, that is, situations where there is no certain regularity, the image information of a set part in multiple consecutive frames of the current camera image is used to judge the movement of the second device and the current first device, and the estimated route is evaluated by the movement situation, so as to accurately calculate and evaluate the movement routes of the current first device and the second device or construction personnel and the possible overlapping areas.

[0070] It should be noted that when obtaining the movement route of the second device or the construction personnel, the movement of the current first device needs to be taken into account. When calculating the real lateral distance and longitudinal distance of the second device or the construction personnel, including the speed, etc., from the relative position change, the movement of the current first device needs to be eliminated.

[0071] Optionally, in the case where the target image is a construction worker, the set parts include feet, legs and upper body; the estimated route of the target image is determined based on image information of the set parts of the target image in multiple consecutive frames of current camera images, including: comparing the set parts with a preset walking image to identify the moving direction of the target image; obtaining the relative position between the set parts and the current first device; calculating the moving speed based on the change in the relative position between the set parts and the current first device in multiple consecutive frames of current camera images; and evaluating the estimated route of the construction worker based on the moving direction and moving speed.

[0072] Optionally, when the target image is a second device, the set part includes the endpoint pixel of the target image that is closest to the current first device; the estimated route of the target image is determined based on the image information of the set part of the target image in multiple consecutive frames of current camera images, including: evaluating the travel direction and moving speed of the second device based on the change in the relative position between the set part and the current first device in multiple consecutive frames of current camera images.

[0073] Taking into account special circumstances such as the displacement of equipment or construction personnel, that is, when there is no certain regularity, the direction and speed of the equipment are relatively stable, while the direction and speed of personnel may change greatly. Therefore, people and equipment are judged separately.

[0074] When a person walks, the distance between the two feet, the direction of the feet, the bending of the legs, the inclination of the upper body and the inclination direction can all be used as typical features to determine the direction of the person's walking. The typical features of the set parts of the human body are combined with the continuous multi-frame current camera images to determine the estimated route of the construction personnel in the next period of time. When the equipment is moving, since the overall movement process of the equipment is relatively stable, the endpoint pixel of the target image closest to the current first device, that is, the point of possible collision, is selected for judgment.

[0075] In this way, by calculating the movement of personnel and equipment separately and comparing and evaluating them through their respective typical parts, the entire judgment process can be made more accurate.

[0076] When the camera captures the construction worker or the second device, it indicates that the construction worker or the second device is relatively close to the first device. Therefore, when evaluating the estimated route of the construction worker or the second device, it is only necessary to evaluate an approximate evaluation route in a short time.

[0077] For example, for construction workers, although the direction and speed of their movement may vary greatly, they still have walking habits. When people walk to the target location, they usually walk in a straight line or in a curve. Therefore, based on the walking habits of people, the estimated route of people is estimated according to the actual route and typical characteristics. Specifically, a route corresponding to the current camera image of multiple consecutive frames is obtained, and the route is matched with the spline curve, that is, it is determined whether the route is a curve or a segment of approximate straight line or multiple segments of approximate straight lines or a curve.

[0078] In the case of an approximate straight line, it indicates that the construction workers are accustomed to walking in a straight line. In this case, the walking direction determined by the typical features of the last frame of the image is used as the walking direction, and the evaluation route of the current construction worker in the case of walking in a straight line is calculated based on the speed.

[0079] In the case of multiple approximate straight lines, it indicates that the current construction workers are accustomed to walking in a straight line but are not moving in a straight line towards the target position. Therefore, since only a short evaluation route needs to be evaluated, the walking direction determined by the typical features of the last frame image is used as the walking direction, and the evaluation route of the current construction workers in the case of walking in a straight line is calculated according to the speed.

[0080] In the case of a curve, the default target position of the construction personnel is in the left front or right front, that is, the route that does not turn back. Draw multiple continuous tangents of the curve from the starting point to the end point of the curve, and obtain the angle change of the curve tangent. If the angle change of the two adjacent curve tangents before the end point becomes smaller and smaller until it is 0 and the angle change is less than or equal to the set a value, the direction of the construction personnel walking towards the target position is basically adjusted. Therefore, the walking direction determined according to the typical features of the last frame image is used as the walking direction, and the evaluation route of the current construction personnel in the case of walking in a straight line is calculated according to the speed; if the angle change of the two adjacent curve tangents before the end point becomes smaller and smaller but the angle change is greater than the set a value or if the angle change of the two adjacent curve tangents before the end point becomes larger and larger, the direction deviation of the construction personnel walking towards the target position is large, indicating that the construction personnel are still adjusting the route. In line with the principle of not turning back, according to the angle change, the starting route direction is used as the initial direction, the direction perpendicular to the starting route direction is used as the final direction, and the estimated curve is fitted according to the actual route, and the estimated curve as a whole or one section of it is used as the evaluation route.

[0081] For the second device, since the moving direction and speed of the device are relatively stable, the route can be evaluated according to the straight line and the direction obtained by the last frame of image.

[0082] Optionally, the danger zone includes the area where the current first device is located, the first danger zone and the second danger zone, the first danger zone is close to the area where the current first device is located, and the second danger zone is outside the first danger zone; in case of overlap, an alarm is simultaneously issued to the current first device and the second device or the construction personnel, including: in case of overlap with the second danger zone, the alarm issues a flat-tone alarm; in case of overlap with the first danger zone, the alarm issues a shifted-tone alarm and a voice alarm to remind the location of the target that is about to collide.

[0083] In the process of collision, the second danger zone is passed first, indicating that the distance to the current first device is relatively close. A collision will occur without changing the moving direction, but there is still a large reaction time to avoid it, so the alarm issued is a flat alarm as a reminder. When passing through the second danger zone and reaching the first danger zone, it indicates that the distance to the current first device is close and needs to be responded immediately, which is quite dangerous, so a variable tone alarm is selected and a voice alarm is issued to remind, and at the same time remind the location of the impending collision, so as to urge the driver and the second device or construction personnel who may collide to respond immediately.

[0084] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operation may vary. The parts and features of some embodiments may be included in or replace the parts and features of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of listings containing one or more associated ones. In addition, when used in the present application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the presence of other identical elements in the process, method or device comprising the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the various embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0085] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. The technicians may use different methods for each specific application to implement the described functions, but such implementations should not be considered to exceed the scope of the embodiments of the present disclosure. The technicians may clearly understand that, for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above may refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.

[0086] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A road construction anti-collision warning method based on visual recognition, comprising: For all operating equipment and construction personnel at the construction site, sensors are installed on the equipment and cameras are set up around the equipment; The construction personnel are provided with sensors; the method comprises: When the camera captures the second device or the construction worker, the relative position between the second device or the construction worker and the current first device is identified according to the current camera image, wherein the relative position includes a lateral distance and a longitudinal distance between the second device or the construction worker and the current first device; Obtaining route information of the current first device and the second device or the construction personnel in the sensor; estimating the estimated routes of the current first device and the second device or the construction personnel according to the route information; Determine whether the estimated route of the second device or the construction personnel overlaps with the dangerous area on the estimated route of the current first device; In the event of overlap, an alarm is simultaneously issued to the current first device and the second device or the construction personnel.

2. The road construction anti-collision warning method based on visual recognition according to claim 1 is characterized in that: The running device is provided with a reference object with scale and direction and can be photographed by the camera. After the camera is installed, the camera is calibrated to obtain the camera internal parameters; when the camera photographs the second device or the construction personnel, the relative position between the second device or the construction personnel and the current first device is identified according to the camera image, including: The distortion of the reference object image captured is corrected according to the camera internal parameters and the reference object, and the corresponding relationship between the pixel coordinates of the corrected image and the scale and direction of the reference object is generated; Correct the current camera image according to the correspondence between the camera internal parameters and the pixel coordinates of the calibrated image and the scale and direction of the reference object; Acquire a target image of the corrected current camera image, where the target image includes a second device or a construction worker; The relative position between the target image and the current first device is calculated according to the target image and the reference object.

3. The method according to claim 2, characterized in that Calculating a relative position between the target image and the current first device according to the target image and the reference object includes: Segment the target image and the camera image according to a preset segmentation model, and generate the edge of the target image; Identify the left and right endpoint pixels of the edge of the target image and record them; Determine the endpoint pixel closest to the current first device among the left and right endpoint pixels according to the reference object; The horizontal distance and the vertical distance between the nearest endpoint pixel and the current first device are calculated according to the reference object.

4. The method according to claim 1, characterized in that The estimated routes of the current first device and the second device or the construction personnel are estimated according to the route information, including: Identify the current working route distribution and cycle of the first device and the second device or the construction personnel and the corresponding time according to the route information; Evaluate the current working scope of the first equipment and the second equipment or the construction personnel according to the distribution of the working routes; Recording the approximate overlapping area of ​​the working range of the current first equipment and the second equipment or the construction personnel within the corresponding time period, wherein the approximate overlapping area includes a completely overlapping area and a preset approximate danger area; The period information of the approximately overlapping area is recorded, and the routes and arrival times of the current first device and the second device or the construction personnel are estimated according to the period information of the approximately overlapping area.

5. The method according to claim 4, characterized in that Estimating the estimated routes of the current first device and the second device or the construction personnel according to the route information, further comprising: Acquire multiple frames of current camera images continuously without approximately overlapping areas; Segmenting a target image in each frame of the current camera image according to a preset segmentation algorithm, wherein the target image includes the second device or the construction personnel; Obtain image information of a set part in a target image; The estimated route of the target image is determined based on the image information of the set part of the target image in multiple consecutive frames of current camera images.

6. The method according to claim 5, characterized in that In the case where the target image is a construction worker, the set parts include feet, legs and upper body; the step of determining the estimated route of the target image based on the image information of the set parts of the target image in the continuous multiple frames of current camera images includes: Compare the set part with the preset walking image to identify the moving direction of the target image; Obtaining the relative position between the set part and the current first device; Calculating the moving speed according to the change in relative position between the set part and the current first device in multiple consecutive frames of current camera images; Evaluate the estimated routes of construction crews based on direction of travel and speed of movement.

7. The method according to claim 5, characterized in that When the target image is a second device, the set part includes the endpoint pixel of the target image that is closest to the current first device; the method of determining the estimated route of the target image based on the image information of the set part of the target image in multiple consecutive frames of current camera images includes: evaluating the travel direction and moving speed of the second device based on the change in the relative position between the set part and the current first device in multiple consecutive frames of current camera images.

8. The method according to claim 1, characterized in that: The danger zone includes the area where the current first device is located, the first danger zone and the second danger zone, the first danger zone is close to the area where the current first device is located, and the second danger zone is outside the first danger zone; in the case of overlap, an alarm is issued to the current first device and the second device or the construction personnel at the same time, including: In case of overlap with the second danger zone, the alarm will sound a level-shifted alarm; In the case of overlap with the first danger zone, the alarm emits a modified tone alarm and a voice alarm to remind the location of the target that is about to collide.

Images