CN118134970A - Jack-up and lifting hook detection tracking method and system based on image recognition - Google Patents
Jack-up and lifting hook detection tracking method and system based on image recognition Download PDFInfo
- Publication number
- CN118134970A CN118134970A CN202410544600.8A CN202410544600A CN118134970A CN 118134970 A CN118134970 A CN 118134970A CN 202410544600 A CN202410544600 A CN 202410544600A CN 118134970 A CN118134970 A CN 118134970A
- Authority
- CN
- China
- Prior art keywords
- detection
- tracking
- crane
- image
- lifting hook
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 328
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000012549 training Methods 0.000 claims description 53
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 11
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 8
- 239000003086 colorant Substances 0.000 description 7
- 238000002372 labelling Methods 0.000 description 5
- 238000005457 optimization Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 238000011478 gradient descent method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a crane and lifting hook detection tracking method based on image recognition, which relates to the field of cranes and comprises the following steps: acquiring a video image of a crane operation site; respectively inputting the images into a personnel detection model and a lifting hook detection model to obtain personnel detection results and lifting hook detection results; determining whether a crane and a lifting hook exist in the detection result; if the crane exists, and the corresponding detection confidence coefficient is greater than or equal to a preset confidence coefficient threshold value, tracking and detecting according to a first set tracking strategy; if the crane exists and the corresponding detection confidence coefficient is smaller than a preset confidence coefficient threshold value, tracking and detecting according to a second set tracking strategy; if no crane exists, tracking and detecting according to a third set tracking strategy; if the lifting hook exists, tracking and detecting according to a fourth set tracking strategy; and if the lifting hook does not exist, tracking and detecting according to a fifth set tracking strategy. The invention can improve the detection precision of the crane and the lifting hook and realize the tracking and monitoring of the states of the crane and the lifting hook.
Description
Technical Field
The invention relates to the technical field of cranes, in particular to a crane and lifting hook detection tracking method and system based on image recognition.
Background
The crane belongs to large-scale mechanical equipment, and in order to ensure the safe operation of the crane, the lifting hook of the crane and a crane for operating the lifting hook and the heavy object need to be tracked in real time in the operation process of the crane.
At present, a target tracking method commonly used in the industrial field adopts a preset tracking target, and the target is tracked in real time through a target tracking algorithm of template, feature matching and mean shift in the target moving process.
However, since the operation site where the crane is located generally has more mechanical devices and operators, and the light of the operation site is generally poor, if the target tracking method is adopted, the detection accuracy is low, and the robustness is poor.
Disclosure of Invention
In order to solve part or all of the technical problems in the prior art, the invention provides a method and a system for detecting and tracking a crane and a lifting hook based on image recognition.
The technical scheme of the invention is as follows:
in a first aspect, there is provided an image recognition-based method for detecting and tracking a crane and a hook, including:
Acquiring a video image of a crane operation site at a tracking moment;
Respectively inputting the video image into a pre-trained personnel detection model and a hook detection model to obtain personnel detection results corresponding to the video image and hook detection results corresponding to the video image;
determining whether a person detected in the person detection result has a crane and whether the lifting hook detected in the lifting hook detection result has a lifting hook;
If the crane exists, and the detection confidence coefficient corresponding to the crane is greater than or equal to a first preset confidence coefficient threshold value, continuous tracking detection of the crane is carried out according to a first set tracking strategy;
If the crane exists, and the detection confidence coefficient corresponding to the crane is smaller than the first preset confidence coefficient threshold value, carrying out continuous tracking detection on the crane according to a second set tracking strategy;
if no crane exists, continuous tracking detection is carried out according to a third set tracking strategy;
If the lifting hook exists, continuous tracking detection of the lifting hook is carried out according to a fourth set tracking strategy;
And if the lifting hook does not exist, carrying out continuous tracking detection according to a fifth set tracking strategy.
In some possible implementations, continuous tracking detection of the crane according to a first set tracking policy includes:
Acquiring images including a current crane in a first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating the duty ratio of the crane in the detection results of all frames of images;
if the duty ratio is smaller than or equal to a first preset ratio, stopping tracking detection;
If the duty ratio is larger than the first preset ratio and smaller than the second preset ratio, the position and the duty ratio of the crane in the image are adjusted, the image including the current crane in the first preset time after the continuous acquisition is carried out, each frame of image acquired by the personnel detection model is detected in real time, the proportion of the crane in the detection result of all frames of images is calculated, if the proportion is smaller than a first preset proportion threshold value, the tracking detection is stopped, and if the proportion is larger than or equal to the first preset proportion threshold value, the step of continuously acquiring the image including the current crane in the first preset time after the continuous acquisition is returned to carry out the continuous tracking detection;
if the duty ratio is greater than or equal to the second preset ratio, returning to the step of acquiring the image including the current crane in the first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating the duty ratio of the crane in the detection results of all frames of images so as to perform continuous tracking detection.
In some possible implementations, continuous tracking detection of the crane according to a second set tracking policy includes:
and processing the video image by using a color channel difference algorithm, and if the current person in the processing result is also a crane, carrying out continuous tracking detection of the current crane according to a first set tracking strategy.
In some possible implementations, processing the video image using a color channel differential algorithm includes:
separating the video image into R, G, B three-channel gray scale images;
respectively calculating and obtaining a gray image obtained by subtracting the gray image of the R channel from the gray image of the B channel, subtracting the gray image of the G channel from the gray image of the R channel, and subtracting the gray image of the R channel from the gray image of the B channel;
Calculating a segmentation threshold value by using a maximum inter-class variance method;
converting the acquired gray level map into a binary map by utilizing a segmentation threshold value;
and judging whether a crane exists according to the acquired binary image.
In some possible implementations, the continuous tracking detection according to the third set tracking policy includes:
acquiring images of a crane operation site in a first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating detection results of all frames of images;
if no lifting tool exists in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting tool in the images, continuously acquiring the images including the current lifting tool in a first preset time, detecting each acquired frame image in real time by using the personnel detection model, calculating the proportion of the lifting tool in the detection results of all the frame images, if the proportion is smaller than a first preset proportion threshold value, stopping tracking detection, and if the proportion is larger than or equal to the first preset proportion threshold value, returning to the step of continuously acquiring the images including the current lifting tool in the first preset time to perform continuous tracking detection.
In some possible implementations, continuous tracking detection of the hook according to a fourth set tracking policy includes:
acquiring images including lifting hooks in a first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating the proportion of the lifting hooks in the detection results of all frames of images;
if the duty ratio is smaller than or equal to a third preset ratio, stopping tracking detection;
If the duty ratio is larger than the third preset ratio and smaller than the fourth preset ratio, the position and the duty ratio of the lifting hook in the image are adjusted, the image including the current lifting hook in the first preset time after the continuous acquisition is carried out, each frame of image acquired by utilizing the lifting hook detection model is detected in real time, the proportion of the lifting hook in the detection results of all frames of images is calculated, if the proportion is smaller than the second preset proportion threshold, the tracking detection is stopped, and if the proportion is larger than or equal to the second preset proportion threshold, the step of continuously acquiring the image including the current lifting hook in the first preset time after the continuous acquisition is returned to carry out the continuous tracking detection;
if the duty ratio is greater than or equal to the fourth preset ratio, returning to the step of acquiring the image including the lifting hook in the first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating the duty ratio of the lifting hook in the detection results of all frames of images so as to perform continuous tracking detection.
In some possible implementations, the continuous tracking detection according to the fifth set tracking policy includes:
Acquiring images of a crane operation site in a first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating detection results of all frames of images;
If the lifting hook does not exist in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting hook in the images, continuously acquiring the images including the current lifting hook in a first preset time, detecting each frame of the acquired images in real time by utilizing the lifting hook detection model, calculating the proportion of the lifting hook existing in the detection results of all the frame images, if the proportion is smaller than a second preset proportion threshold value, stopping tracking detection, and if the proportion is greater than or equal to the second preset proportion threshold value, returning to the step of continuously acquiring the images including the current lifting hook in the first preset time after the continuous acquisition so as to perform continuous tracking detection.
In some possible implementations, the person detection model is trained by:
acquiring a first training sample set, wherein the first training sample comprises a sample image, a crane corresponding to the sample image and the position of the crane;
And taking a sample image in the first training sample set as input, and taking a crane and the position thereof corresponding to the input sample image as output to train the personnel detection model.
In some possible implementations, the hook detection model is trained by:
acquiring a second training sample set, wherein the second training sample comprises a sample image, a lifting hook corresponding to the sample image and the position of the lifting hook;
And taking a sample image in the second training sample set as input, taking a lifting hook corresponding to the input sample image and the position thereof as output, and training the lifting hook detection model.
In a second aspect, there is also provided an image recognition-based lifting jack and hook detection tracking system, comprising:
the image acquisition module is configured to acquire a video image of a crane operation site;
the detection module is configured to input the video image into a pre-trained personnel detection model and a hook detection model respectively to obtain personnel detection results corresponding to the video image and hook detection results corresponding to the video image;
a detection result determining module configured to determine whether a person detected in the person detection result has a crane, and whether the hook detected in the hook detection result has a hook;
the first tracking detection control module is configured to perform continuous tracking detection of the crane according to a first set tracking strategy when the crane exists and the detection confidence corresponding to the crane is greater than or equal to a first preset confidence threshold;
the second tracking detection control module is configured to perform continuous tracking detection of the crane according to a second set tracking strategy when the crane exists and the detection confidence corresponding to the crane is smaller than a first preset confidence threshold;
The third tracking detection control module is configured to perform continuous tracking detection according to a third set tracking strategy when no lifting exists;
The fourth tracking detection control module is configured to perform continuous tracking detection of the lifting hook according to a fourth set tracking strategy when the lifting hook exists;
And the fifth tracking detection control module is configured to perform continuous tracking detection according to a fifth set tracking strategy when the lifting hook is not present.
The technical scheme of the invention has the main advantages that:
According to the method and the system for detecting and tracking the lifting hook based on the image identification, the image of the crane operation site is detected by using the detection model, and the lifting hook are tracked and detected by adopting different tracking strategies according to different detection results, so that the detection precision of the lifting hook and the lifting hook can be improved, the stable tracking and monitoring of the working states of the lifting hook and the lifting hook can be realized, and the safety of the lifting operation can be ensured.
Drawings
The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention and without limitation to the invention. In the drawings:
FIG. 1 is a flow chart of a method for detecting and tracking a crane and a hook based on image recognition according to an embodiment of the invention;
FIG. 2 is a flowchart of a first configuration tracking strategy according to an embodiment of the present invention;
FIG. 3 is a flowchart of a third configuration tracking strategy according to an embodiment of the present invention;
FIG. 4 is a flowchart of a fourth configuration tracking strategy according to an embodiment of the present invention;
FIG. 5 is a flowchart of a fifth configuration tracking strategy according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to specific embodiments of the present invention and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The following describes in detail the technical scheme provided by the embodiment of the invention with reference to the accompanying drawings.
Referring to fig. 1, in a first aspect, an embodiment of the present invention provides a method for detecting and tracking a crane and a hook based on image recognition, the method comprising the steps of S1 to S8:
Step S1, acquiring video images of a crane operation site at tracking time.
In one embodiment of the invention, a camera is installed on a crane operation site, and a video image of the crane operation site at a designated initial tracking time is acquired by the camera.
In view of the fact that a plurality of cranes may be provided at the hoisting site, for this purpose, when a plurality of cranes are provided, a plurality of cameras are installed at the hoisting site, each for capturing video images of a plurality of different crane working sites.
And S2, respectively inputting the video image into a pre-trained personnel detection model and a pre-trained lifting hook detection model to obtain personnel detection results corresponding to the video image and lifting hook detection results corresponding to the video image.
Specifically, the video image is input into a pre-trained personnel detection model, the personnel detection model outputs personnel detection results corresponding to the video image, the video image is input into a pre-trained lifting hook detection model, and the lifting hook detection results corresponding to the video image are output by the lifting hook detection model.
Further, in one embodiment of the present invention, the human detection model is trained by:
acquiring a first training sample set, wherein the first training sample comprises a sample image, a crane corresponding to the sample image and the position of the crane;
and taking a sample image in a first training sample in the first training sample set as input, taking a crane corresponding to the input sample image and the position thereof as output, and training a personnel detection model.
In an embodiment of the present invention, the sample image may be selected from existing images, or may be obtained by shooting with a camera. The number of the sample images in the first training sample set is determined according to actual requirements, and in general, the greater the number of the sample images, the higher the accuracy of the trained detection model, but the higher the required sample image acquisition cost and training cost.
In consideration of that a plurality of workers are arranged on a lifting site, the workers are generally classified into a plurality of different work types, in order to facilitate detection of the lifting workers and improve detection accuracy, in the practical application process, the lifting workers can wear safety helmets with specific colors, personnel of other work types wear safety helmets with other colors, for example, the lifting workers wear white safety helmets, other personnel wear safety helmets with other colors such as blue, red, yellow, orange and the like, at the moment, lifting worker labels in sample images are corresponding safety helmet labels with specific colors, and detection of the lifting workers is carried out through the safety helmets worn by detection personnel.
Further, in an embodiment of the present invention, a sample image in a first training sample set is taken as an input, a crane and a position thereof corresponding to the input sample image are taken as an output, and a training person detects a model, and further includes the following steps:
Step S201, sequentially inputting sample images in each first training sample in the first training sample set into a personnel detection model to obtain a prediction crane and a position thereof output by the personnel detection model.
In an embodiment of the invention, a sample image in a first training sample is input from an input end of a personnel detection model, sequentially processed by parameters of each layer in the personnel detection model, and output from an output end of the personnel detection model, wherein information output by the output end of the personnel detection model is a prediction crane corresponding to the sample image and the position of the prediction crane.
In an embodiment of the present invention, the person detection model is a model to be trained, parameters of each layer of the model are initialization parameters, and parameters of each layer of the model are continuously updated in the training process of the model.
Step S202, calculating a preset target detection loss function according to the crane and the position thereof corresponding to the sample image in each first training sample and the predicted crane and the position thereof corresponding to the sample image and output by the personnel detection model.
In an embodiment of the present invention, the objective detection loss function is specifically set according to the actual situation. For example, a conventional target detection loss function is adopted, specifically expressed as:
；
Wherein, Representing classification loss,/>Indicating a loss of position. The classification Loss and the location Loss may employ conventional Loss functions, for example, the classification Loss employs a cross entropy Loss function, the location Loss employs a Smooth L1 Loss or an IOU Loss, or the like.
Specifically, based on the set target detection loss function, the target detection loss function is calculated from the crane and its position corresponding to the sample image in each first training sample, and the predicted crane and its position corresponding to the sample image output by the human detection model.
Step S203, judging whether a preset training stop condition is reached, if yes, taking the current personnel detection model as a personnel detection model for completing training, if not, updating parameters of the personnel detection model by using a preset target detection loss function, and returning to the step S201.
In an embodiment of the present invention, the training stop condition is specifically set according to the actual situation, for example, the training iteration number reaches a set iteration algebra; or the target detection loss function is used as an optimization index of the personnel detection model, and the training stopping condition is set to be that the optimization index reaches a set threshold.
In one embodiment of the invention, a random gradient descent method is adopted to train and update parameters of the personnel detection model.
Further, in one embodiment of the present invention, the hook detection model is trained by:
acquiring a second training sample set, wherein the second training sample comprises a sample image, a lifting hook corresponding to the sample image and the position of the lifting hook;
And taking a sample image in a second training sample in the second training sample set as input, taking a lifting hook corresponding to the input sample image and the position thereof as output, and training a lifting hook detection model.
Further, in an embodiment of the present invention, a sample image in a second training sample in the second training sample set is taken as an input, a hook and a position thereof corresponding to the input sample image are taken as an output, and a hook detection model is trained, and the method further includes the following steps:
Step S211, sequentially inputting sample images in each second training sample in the second training sample set into the hook detection model to obtain a predicted hook and the position thereof output by the hook detection model.
In an embodiment of the invention, a sample image in a second training sample is input from an input end of a hook detection model, sequentially processed by parameters of each layer in the hook detection model, and output from an output end of the hook detection model, wherein the information output by the output end of the hook detection model is the predicted hook and the position thereof corresponding to the sample image.
In an embodiment of the present invention, the hook detection model is a model to be trained, parameters of each layer of the model are initialization parameters, and parameters of each layer of the model are continuously updated during the training process of the model.
Step S212, calculating a preset target detection loss function according to the lifting hook and the position thereof corresponding to the sample image in each second training sample and the predicted lifting hook and the position thereof corresponding to the sample image and output by the lifting hook detection model.
In an embodiment of the present invention, the objective detection loss function is specifically set according to the actual situation. For example, a conventional target detection loss function is adopted, specifically expressed as:
；
Wherein, Representing classification loss,/>Indicating a loss of position. The classification Loss and the location Loss may employ conventional Loss functions, for example, the classification Loss employs a cross entropy Loss function, the location Loss employs a Smooth L1 Loss or an IOU Loss, or the like.
Specifically, based on the set target detection loss function, the target detection loss function is calculated according to the hooks and positions thereof corresponding to the sample images in each second training sample, and the predicted hooks and positions thereof corresponding to the sample images and output by the hook detection model.
Step S213, judging whether a preset training stopping condition is reached, if yes, taking the current lifting hook detection model as a lifting hook detection model for completing training, if not, updating parameters of the lifting hook detection model by using a preset target detection loss function, and returning to step S211.
In an embodiment of the present invention, the training stop condition is specifically set according to the actual situation, for example, the training iteration number reaches a set iteration algebra; or the target detection loss function is used as an optimization index of the lifting hook detection model, and the training stopping condition is set to be that the optimization index reaches a set threshold.
In an embodiment of the present invention, a random gradient descent method is used to perform training and updating of parameters of the hook detection model.
In an embodiment of the present invention, the personnel detection model and the hook detection model both use YOLO series network models.
Further, as the number of sample images is large, and the detection model needs to be optimized iteratively, the time consumed for labeling the labels such as the crane and the lifting hook one by one is long. In order to improve label labeling efficiency, in an embodiment of the invention, labels of samples are obtained by setting each label sample to be matched and then according to SIFT feature point extraction and RANSAC optimization optimal matching algorithm, batch labeling of sample images is completed, whether labeled features are corresponding labels is finally confirmed, and batch labeling of the labels of the sample images is performed in a processing mode of completing the whole feature labeling process.
Further, in order to improve the detection accuracy and reduce false detection, the width of the crane feature and the width of the hook feature in the sample image are not smaller than 96 pixels.
And step S3, determining whether a person detected in the person detection result has a crane and whether the lifting hook detected in the lifting hook detection result has a lifting hook.
Specifically, whether the crane is present or not and whether the hook is present or not are determined according to the detection result obtained in step S2.
And S4, if the crane exists, and the detection confidence coefficient corresponding to the crane is greater than or equal to a first preset confidence coefficient threshold value, carrying out continuous tracking detection on the crane according to a first set tracking strategy.
Referring to fig. 2, in order to ensure stable tracking detection of a crane in view of complex environment and generally poor line of sight of a crane site, in an embodiment of the present invention, continuous tracking detection of a crane is performed according to a first set tracking strategy, including the steps of:
Step S401, acquiring images including a current crane in a first preset time after the current moment, detecting each acquired frame of images in real time by using a personnel detection model, and calculating the duty ratio of the crane in the detection results of all frames of images;
step S402, if the duty ratio is smaller than or equal to a first preset ratio, stopping tracking detection;
Step S403, if the duty ratio is larger than the first preset ratio and smaller than the second preset ratio, the position and the duty ratio of the crane in the image are adjusted, the image including the current crane in the first preset time after the continuous acquisition is carried out, each frame of image acquired by the real-time detection of the personnel detection model is utilized, the proportion of the crane in the detection result of all frames of images is calculated, if the proportion is smaller than the first preset proportion threshold, the tracking detection is stopped, and if the proportion is larger than or equal to the first preset proportion threshold, the step 'the image including the current crane in the first preset time after the continuous acquisition' is returned to carry out the continuous tracking detection;
Step S404, if the duty ratio is greater than or equal to the second preset ratio, returning to step S401 for continuous tracking detection.
In one embodiment of the present invention, in step S403, the position and the duty ratio of the crane in the image are adjusted so that the position of the crane feature in the image is centered and the duty ratio is within the set ratio range.
Specifically, in one embodiment of the present invention, the position and duty cycle of the crane in the image are adjusted to center the position of the crane feature in the image and the duty cycle is within a set ratio range by:
If the position of the crane characteristic is deviated left or right relative to the center of the image, according to the relation between the pixel coordinates of the image and the view angle parameter of the camera, obtaining the angle of the camera to be rotated and controlling the rotation of the camera, so that the crane characteristic is in the position in the middle of the image; when the crane feature is centered in the image, if the crane feature is smaller, the focal length is adaptively enlarged so that the area occupied by the crane feature in the image is within a set ratio range, and if the crane feature is larger, the focal length is adaptively reduced so that the area occupied by the crane feature in the image is within the set ratio range.
Further, the rotation angle of the camera is determined by:
Assuming that the field angle of the camera is VFOV (vertical field angle) and HFOV (horizontal field angle), the parameters are camera parameters, the coordinates of the crane feature on the image are (x, y), the resolution of the image is (w, h), and the relative position of the crane feature on the image, that is, the offset of the crane feature from the center point of the image is calculated. Assuming that the center point of the image is (cx, cy), the relative position of the crane feature on the image is (dx, dy), where dx=x-cx, dy=y-cy. And calculating the horizontal angle and the inclined angle of the rotation of the cradle head according to the relative position of the crane characteristic on the image and the field angle of the camera. Assuming that the relative positions of the crane features on the image are dx and dy, respectively, the horizontal angle of rotation is: a=dx/w HFOV, the tilt angle of rotation is: b=dy/h VFOV, where w is the width of the image and h is the height of the image.
And S5, if the crane exists, and the detection confidence coefficient corresponding to the crane is smaller than the first preset confidence coefficient threshold value, carrying out continuous tracking detection on the crane according to a second set tracking strategy.
In order to ensure stable tracking detection of the crane, in one embodiment of the present invention, continuous tracking detection of the crane is performed according to a second set tracking strategy, which includes the following steps:
And processing the video image by utilizing a color channel differential algorithm, and if the current person is a crane in the processing result, carrying out continuous tracking detection of the current crane according to a first set tracking strategy.
Specifically, the method for processing the video image by utilizing the color channel difference algorithm comprises the following steps:
step S501, separating a video image into R, G, B three-channel gray scale images;
Step S502, respectively calculating and obtaining a gray image obtained by subtracting the gray image of the R channel from the gray image of the B channel, subtracting the gray image of the G channel from the gray image of the R channel, and subtracting the gray image of the R channel from the gray image of the B channel;
Step S503, calculating a segmentation threshold value by using a maximum inter-class variance method (OTSU algorithm);
Step S504, converting the acquired gray level map into a binary map by using a segmentation threshold;
Step S505, judging whether a crane exists according to the acquired binary image.
Because R, G, B channel values of different colors of the composite color are different, and R, G, B values of the same color under different light rays and different backgrounds are different due to inconsistent light ray intensity, in one embodiment of the invention, the color can be accurately distinguished by processing the video image through the color channel difference algorithm, and whether a crane wearing a safety helmet with a specific color exists in the video image is determined.
Further, since the crane feature has a certain duty ratio in the image, in order to determine the appropriate segmentation threshold, in step S503, calculating the segmentation threshold using the maximum inter-class variance method further includes:
Firstly, carrying out maximum inter-class variance calculation on a gray level diagram to obtain a segmentation threshold;
and calculating the percentage ratio of the acquired segmentation threshold, and circularly increasing or decreasing the segmentation threshold by a step length of 1 according to the percentage ratio until the segmentation threshold meeting the specified percentage is obtained.
Wherein, the appointed percentage is specifically determined according to the actual situation.
Further, in step S505, determining whether a crane exists according to the acquired binary image specifically includes:
And calculating the number and the size of the connected domains of the current image, calculating the length-width ratio of the connected domains, judging whether the connected domains are the characteristics of the crane according to the morphological and dimensional parameters of the connected domains and the corresponding colors, if so, determining that the crane exists, and if not, determining that the crane does not exist.
For example, taking the above-defined white safety helmet as an example, if it is determined that the white safety helmet exists according to the form and size parameters of the connected domain and the corresponding colors, the presence of the crane is indicated.
And S6, if no crane exists, continuous tracking detection is carried out according to a third set tracking strategy.
Referring to fig. 3, in order to ensure stable tracking detection of a crane, in an embodiment of the present invention, continuous tracking detection is performed according to a third set tracking strategy, considering that the environment of the crane site is complex and the line of sight is generally poor, including the following steps:
step S601, acquiring images of a crane operation site within a first preset time after the current moment, detecting each acquired frame of images in real time by using a personnel detection model, and calculating detection results of all frames of images;
Step S602, if no lifting jack exists in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting jack in the images, continuously acquiring the images including the current lifting jack in a first preset time, detecting each frame image acquired by utilizing a personnel detection model in real time, calculating the proportion of the lifting jack in the detection results of all the frame images, if the proportion is smaller than a first preset proportion threshold, stopping tracking detection, if the proportion is greater than or equal to the first preset proportion threshold, and returning to the step of continuously acquiring the images including the current lifting jack in the first preset time to perform continuous tracking detection.
And S7, if the lifting hook exists, carrying out continuous tracking detection of the lifting hook according to a fourth set tracking strategy.
Referring to fig. 4, in order to ensure stable tracking detection of the hook in consideration of complex environment and poor line of sight of the lifting site, in an embodiment of the present invention, continuous tracking detection of the hook is performed according to a fourth set tracking strategy, which includes the following steps:
step S701, acquiring images including hooks in a first preset time after the current moment, detecting each acquired frame of image in real time by using a hook detection model, and calculating the duty ratio of hooks in detection results of all frames of images;
Step S702, if the duty ratio is smaller than or equal to a third preset ratio, stopping tracking detection;
step S703, if the duty ratio is greater than the third preset ratio and less than the fourth preset ratio, adjusting the position and duty ratio of the hook in the image, continuously acquiring the image including the current hook in the first preset time after the image is continuously acquired, detecting each acquired frame of image in real time by using the hook detection model, calculating the proportion of hooks in the detection results of all frames of images, if the proportion is less than the second preset proportion threshold, stopping tracking detection, and if the proportion is greater than or equal to the second preset proportion threshold, returning to the step of continuously acquiring the image including the current hook in the first preset time after the image is continuously acquired to perform continuous tracking detection;
step S704, if the duty ratio is greater than or equal to the fourth preset ratio, returning to step S701 for continuous tracking detection.
And S8, if the lifting hook does not exist, continuous tracking detection is carried out according to a fifth set tracking strategy.
Referring to fig. 5, in order to ensure stable tracking detection of the hook, in an embodiment of the present invention, continuous tracking detection is performed according to a fifth set tracking strategy, which includes:
Step S801, acquiring images of a crane operation site within a first preset time after the current moment, detecting each acquired frame of images in real time by using a lifting hook detection model, and calculating detection results of all frames of images;
Step S802, if no lifting hook exists in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting hook in the images, continuously acquiring the images including the current lifting hook in a first preset time, detecting each acquired frame image in real time by using a lifting hook detection model, calculating the proportion of the lifting hook in the detection results of all the frame images, if the proportion is smaller than a second preset proportion threshold, stopping tracking detection, and if the proportion is greater than or equal to the second preset proportion threshold, returning to the step 'continuously acquiring the images including the current lifting hook in the first preset time so as to perform continuous tracking detection'.
Further, in an embodiment of the present invention, specific values of the first preset confidence threshold, the first preset time, the first preset ratio, the second preset ratio, the first preset ratio threshold, the set ratio range, the third preset ratio, the fourth preset ratio, and the second preset ratio threshold are specifically set according to actual situations. For example, the first preset confidence threshold is set to 90%, the first preset time is set to 5s, the first preset ratio is set to 25%, the second preset ratio is set to 80%, the first preset ratio threshold is set to 25%, the set ratio range is set to 2-3%, the third preset ratio is set to 25%, the fourth preset ratio is set to 80%, and the second preset ratio threshold is set to 25%.
In a second aspect, an embodiment of the present invention further provides an image recognition-based crane and hook detection tracking system, including:
the image acquisition module is configured to acquire a video image of a crane operation site;
The detection module is configured to input the video image into a pre-trained personnel detection model and a pre-trained lifting hook detection model respectively to obtain personnel detection results corresponding to the video image and lifting hook detection results corresponding to the video image;
the detection result determining module is configured to determine whether a person detected in the person detection result has a crane and whether the lifting hook detected in the lifting hook detection result has a lifting hook;
the first tracking detection control module is configured to perform continuous tracking detection of the crane according to a first set tracking strategy when the crane exists and the detection confidence corresponding to the crane is greater than or equal to a first preset confidence threshold;
the second tracking detection control module is configured to perform continuous tracking detection of the crane according to a second set tracking strategy when the crane exists and the detection confidence corresponding to the crane is smaller than a first preset confidence threshold;
The third tracking detection control module is configured to perform continuous tracking detection according to a third set tracking strategy when no lifting exists;
The fourth tracking detection control module is configured to perform continuous tracking detection of the lifting hook according to a fourth set tracking strategy when the lifting hook exists;
And the fifth tracking detection control module is configured to perform continuous tracking detection according to a fifth set tracking strategy when the lifting hook is not present.
The above modules are devices corresponding to the steps of the above method, and the specific working principle and the beneficial effects of each module can be referred to the above method, which is not described herein again.
According to the method and the system for detecting and tracking the lifting hook based on the image recognition, the image of the crane operation site is detected by using the detection model, and the lifting hook are tracked and detected by using different tracking strategies according to different detection results, so that the detection precision of the lifting hook and the lifting hook can be improved, the stable tracking and monitoring of the working states of the lifting hook and the lifting hook can be realized, and the safety of the lifting operation can be ensured.
It should be noted that in this document, relational terms such as "first" and "second" and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting thereof; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. The method for detecting and tracking the lifting hook and the lifting jack based on the image recognition is characterized by comprising the following steps:
Acquiring a video image of a crane operation site at a tracking moment;
Respectively inputting the video image into a pre-trained personnel detection model and a hook detection model to obtain personnel detection results corresponding to the video image and hook detection results corresponding to the video image;
determining whether a person detected in the person detection result has a crane and whether the lifting hook detected in the lifting hook detection result has a lifting hook;
If the crane exists, and the detection confidence coefficient corresponding to the crane is greater than or equal to a first preset confidence coefficient threshold value, continuous tracking detection of the crane is carried out according to a first set tracking strategy;
If the crane exists, and the detection confidence coefficient corresponding to the crane is smaller than the first preset confidence coefficient threshold value, carrying out continuous tracking detection on the crane according to a second set tracking strategy;
if no crane exists, continuous tracking detection is carried out according to a third set tracking strategy;
If the lifting hook exists, continuous tracking detection of the lifting hook is carried out according to a fourth set tracking strategy;
And if the lifting hook does not exist, carrying out continuous tracking detection according to a fifth set tracking strategy.
2. The image recognition-based lift and hook detection and tracking method of claim 1, wherein the continuous tracking detection of the lift according to the first set tracking strategy comprises:
Acquiring images including a current crane in a first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating the duty ratio of the crane in the detection results of all frames of images;
if the duty ratio is smaller than or equal to a first preset ratio, stopping tracking detection;
If the duty ratio is larger than the first preset ratio and smaller than the second preset ratio, the position and the duty ratio of the crane in the image are adjusted, the image including the current crane in the first preset time after the continuous acquisition is carried out, each frame of image acquired by the personnel detection model is detected in real time, the proportion of the crane in the detection result of all frames of images is calculated, if the proportion is smaller than a first preset proportion threshold value, the tracking detection is stopped, and if the proportion is larger than or equal to the first preset proportion threshold value, the step of continuously acquiring the image including the current crane in the first preset time after the continuous acquisition is returned to carry out the continuous tracking detection;
if the duty ratio is greater than or equal to the second preset ratio, returning to the step of acquiring the image including the current crane in the first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating the duty ratio of the crane in the detection results of all frames of images so as to perform continuous tracking detection.
3. The image recognition-based lift and hook detection and tracking method of claim 2, wherein the continuous tracking detection of the lift according to the second set tracking strategy comprises:
and processing the video image by using a color channel difference algorithm, and if the current person in the processing result is also a crane, carrying out continuous tracking detection of the current crane according to a first set tracking strategy.
4. The image recognition-based lift and hook detection and tracking method of claim 3 wherein processing the video image using a color channel differential algorithm comprises:
separating the video image into R, G, B three-channel gray scale images;
respectively calculating and obtaining a gray image obtained by subtracting the gray image of the R channel from the gray image of the B channel, subtracting the gray image of the G channel from the gray image of the R channel, and subtracting the gray image of the R channel from the gray image of the B channel;
Calculating a segmentation threshold value by using a maximum inter-class variance method;
converting the acquired gray level map into a binary map by utilizing a segmentation threshold value;
and judging whether a crane exists according to the acquired binary image.
5. The image recognition-based lift and hook detection and tracking method of claim 1, wherein the continuous tracking detection according to a third set tracking strategy comprises:
acquiring images of a crane operation site in a first preset time after the current moment, detecting each acquired frame of image in real time by using the personnel detection model, and calculating detection results of all frames of images;
if no lifting tool exists in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting tool in the images, continuously acquiring the images including the current lifting tool in a first preset time, detecting each acquired frame image in real time by using the personnel detection model, calculating the proportion of the lifting tool in the detection results of all the frame images, if the proportion is smaller than a first preset proportion threshold value, stopping tracking detection, and if the proportion is larger than or equal to the first preset proportion threshold value, returning to the step of continuously acquiring the images including the current lifting tool in the first preset time to perform continuous tracking detection.
6. The image recognition-based lift and hook detection and tracking method of claim 1, wherein the continuous tracking detection of the hook according to the fourth set tracking strategy comprises:
acquiring images including lifting hooks in a first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating the proportion of the lifting hooks in the detection results of all frames of images;
if the duty ratio is smaller than or equal to a third preset ratio, stopping tracking detection;
If the duty ratio is larger than the third preset ratio and smaller than the fourth preset ratio, the position and the duty ratio of the lifting hook in the image are adjusted, the image including the current lifting hook in the first preset time after the continuous acquisition is carried out, each frame of image acquired by utilizing the lifting hook detection model is detected in real time, the proportion of the lifting hook in the detection results of all frames of images is calculated, if the proportion is smaller than the second preset proportion threshold, the tracking detection is stopped, and if the proportion is larger than or equal to the second preset proportion threshold, the step of continuously acquiring the image including the current lifting hook in the first preset time after the continuous acquisition is returned to carry out the continuous tracking detection;
if the duty ratio is greater than or equal to the fourth preset ratio, returning to the step of acquiring the image including the lifting hook in the first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating the duty ratio of the lifting hook in the detection results of all frames of images so as to perform continuous tracking detection.
7. The image recognition-based lift and hook detection and tracking method of claim 1, wherein the continuous tracking detection according to a fifth set tracking strategy comprises:
Acquiring images of a crane operation site in a first preset time after the current moment, detecting each acquired frame of image in real time by using the lifting hook detection model, and calculating detection results of all frames of images;
If the lifting hook does not exist in the detection results of all the frame images, stopping tracking detection, otherwise, adjusting the position and the duty ratio of the lifting hook in the images, continuously acquiring the images including the current lifting hook in a first preset time, detecting each frame of the acquired images in real time by utilizing the lifting hook detection model, calculating the proportion of the lifting hook existing in the detection results of all the frame images, if the proportion is smaller than a second preset proportion threshold value, stopping tracking detection, and if the proportion is greater than or equal to the second preset proportion threshold value, returning to the step of continuously acquiring the images including the current lifting hook in the first preset time after the continuous acquisition so as to perform continuous tracking detection.
8. The image recognition-based lift and hook detection and tracking method of claim 1, wherein the personnel detection model is trained by:
acquiring a first training sample set, wherein the first training sample comprises a sample image, a crane corresponding to the sample image and the position of the crane;
And taking a sample image in the first training sample set as input, and taking a crane and the position thereof corresponding to the input sample image as output to train the personnel detection model.
9. The image recognition-based lift and hook detection tracking method of claim 1, wherein the hook detection model is trained by:
acquiring a second training sample set, wherein the second training sample comprises a sample image, a lifting hook corresponding to the sample image and the position of the lifting hook;
And taking a sample image in the second training sample set as input, taking a lifting hook corresponding to the input sample image and the position thereof as output, and training the lifting hook detection model.
10. A crane and hook detection tracking system based on image recognition, comprising:
the image acquisition module is configured to acquire a video image of a crane operation site;
the detection module is configured to input the video image into a pre-trained personnel detection model and a hook detection model respectively to obtain personnel detection results corresponding to the video image and hook detection results corresponding to the video image;
a detection result determining module configured to determine whether a person detected in the person detection result has a crane, and whether the hook detected in the hook detection result has a hook;
the first tracking detection control module is configured to perform continuous tracking detection of the crane according to a first set tracking strategy when the crane exists and the detection confidence corresponding to the crane is greater than or equal to a first preset confidence threshold;
the second tracking detection control module is configured to perform continuous tracking detection of the crane according to a second set tracking strategy when the crane exists and the detection confidence corresponding to the crane is smaller than a first preset confidence threshold;
The third tracking detection control module is configured to perform continuous tracking detection according to a third set tracking strategy when no lifting exists;
The fourth tracking detection control module is configured to perform continuous tracking detection of the lifting hook according to a fourth set tracking strategy when the lifting hook exists;
And the fifth tracking detection control module is configured to perform continuous tracking detection according to a fifth set tracking strategy when the lifting hook is not present.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410544600.8A CN118134970B (en) | 2024-05-06 | 2024-05-06 | Jack-up and lifting hook detection tracking method and system based on image recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410544600.8A CN118134970B (en) | 2024-05-06 | 2024-05-06 | Jack-up and lifting hook detection tracking method and system based on image recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118134970A true CN118134970A (en) | 2024-06-04 |
| CN118134970B CN118134970B (en) | 2024-07-09 |
Family
ID=91238133
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410544600.8A Active CN118134970B (en) | 2024-05-06 | 2024-05-06 | Jack-up and lifting hook detection tracking method and system based on image recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118134970B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060104487A1 (en) * | 2002-11-29 | 2006-05-18 | Porter Robert M S | Face detection and tracking |
| WO2011149145A1 (en) * | 2010-05-24 | 2011-12-01 | 중앙대학교 산학협력단 | Device and method for tracing non-rigid objects using shapes and feature information |
| US20180300553A1 (en) * | 2017-03-30 | 2018-10-18 | Hrl Laboratories, Llc | Neuromorphic system for real-time visual activity recognition |
| CN111127518A (en) * | 2019-12-24 | 2020-05-08 | 深圳火星探索科技有限公司 | Target tracking method and device based on unmanned aerial vehicle |
| CN111242977A (en) * | 2020-01-09 | 2020-06-05 | 影石创新科技股份有限公司 | Target tracking method of panoramic video, readable storage medium and computer equipment |
| KR102160749B1 (en) * | 2019-05-29 | 2020-09-28 | 재단법인대구경북과학기술원 | Method and apparatus for object tracking |
| CN115180522A (en) * | 2022-05-31 | 2022-10-14 | 品茗科技股份有限公司 | Safety monitoring method and system for hoisting device construction site |
| CN116385485A (en) * | 2023-03-13 | 2023-07-04 | 腾晖科技建筑智能(深圳)有限公司 | Video tracking method and system for long-strip-shaped tower crane object |
-
2024
- 2024-05-06 CN CN202410544600.8A patent/CN118134970B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060104487A1 (en) * | 2002-11-29 | 2006-05-18 | Porter Robert M S | Face detection and tracking |
| WO2011149145A1 (en) * | 2010-05-24 | 2011-12-01 | 중앙대학교 산학협력단 | Device and method for tracing non-rigid objects using shapes and feature information |
| US20180300553A1 (en) * | 2017-03-30 | 2018-10-18 | Hrl Laboratories, Llc | Neuromorphic system for real-time visual activity recognition |
| KR102160749B1 (en) * | 2019-05-29 | 2020-09-28 | 재단법인대구경북과학기술원 | Method and apparatus for object tracking |
| CN111127518A (en) * | 2019-12-24 | 2020-05-08 | 深圳火星探索科技有限公司 | Target tracking method and device based on unmanned aerial vehicle |
| CN111242977A (en) * | 2020-01-09 | 2020-06-05 | 影石创新科技股份有限公司 | Target tracking method of panoramic video, readable storage medium and computer equipment |
| CN115180522A (en) * | 2022-05-31 | 2022-10-14 | 品茗科技股份有限公司 | Safety monitoring method and system for hoisting device construction site |
| CN116385485A (en) * | 2023-03-13 | 2023-07-04 | 腾晖科技建筑智能(深圳)有限公司 | Video tracking method and system for long-strip-shaped tower crane object |
Non-Patent Citations (1)
| Title |
|---|
| 武婷: "基于视频的塔吊吊钩自动跟踪算法研究", 《中国优秀硕士学位论文全文数据库》, no. 3, 15 March 2018 (2018-03-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118134970B (en) | 2024-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111898514B (en) | Multi-target visual supervision method based on target detection and action recognition | |
| JP4328286B2 (en) | Face area estimation device, face area estimation method, and face area estimation program | |
| CN109493313B (en) | Vision-based steel coil positioning method and equipment | |
| CN107506760A (en) | Traffic signals detection method and system based on GPS location and visual pattern processing | |
| CN112101260B (en) | Method, device, equipment and storage medium for identifying safety belt of operator | |
| CN114812403B (en) | Large-span steel structure hoisting deformation monitoring method based on unmanned plane and machine vision | |
| CN114842397A (en) | Real-time old man falling detection method based on anomaly detection | |
| CN115205977A (en) | Method and system for identifying unsafe behaviors of subway construction operators | |
| CN118134970B (en) | Jack-up and lifting hook detection tracking method and system based on image recognition | |
| CN107045630A (en) | A kind of pedestrian detection and personal identification method and system based on RGBD | |
| CN110728212A (en) | Road well lid monitoring device and method based on computer vision | |
| CN117474321B (en) | BIM model-based construction site risk intelligent identification method and system | |
| CN109993715A (en) | A kind of robot vision image preprocessing system and image processing method | |
| CN116883661B (en) | Fire operation detection method based on target identification and image processing | |
| CN117787690A (en) | Hoisting operation safety risk identification method and identification device | |
| CN111461079A (en) | Binocular image-based method for detecting personnel under suspension arm | |
| KR102171384B1 (en) | Object recognition system and method using image correction filter | |
| CN115082509B (en) | Method for tracking non-feature target | |
| CN115797411B (en) | Method for online recognition of hydropower station cable bridge deformation by utilizing machine vision | |
| CN116532623A (en) | Slag adding robot obstacle recognition method capable of adapting to strong and weak illumination change | |
| CN110598521A (en) | Behavior and physiological state identification method based on intelligent analysis of face image | |
| CN106339666B (en) | A kind of night monitoring method of human body target | |
| CN112551400A (en) | Tower crane jib loading boom deformation detecting system based on monocular vision detection | |
| CN110533698B (en) | Foundation pit construction pile detection control method based on visual detection | |
| CN111583341A (en) | Pan-tilt camera displacement detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |