Helmet standard wearing identification method and system considering head key point positions

Abstract

The invention relates to the technical field of computer vision security and protection, and discloses a helmet normative wearing identification method and system considering head key point positions, and the method comprises the steps: constructing a human body detection model and a helmet normative wearing classification model, and carrying out the training; collecting a monitoring video, performing frame extraction on a video stream according to a specific time interval to obtain a video frame, and positioning a human body in the video frame by using the trained human body detection model to output a human body image; reasoning each human body image through a human body key point recognition algorithm, positioning a plurality of key point positions of a human body, judging a head deflection posture by taking mutual positions among key points in a head region, and cutting a head region image according to the key points in the head region; and inputting the head region image into the standard helmet wearing classification model for reasoning and recognition, judging the head posture through head key point recognition, and judging whether the helmet is worn in a standard manner through the classification model, so that the method has relatively high generalization ability and recognition precision.

Description

Technical Field

[0001] The present invention relates to the field of computer vision security technology, and in particular to a method and system for identifying the standard wearing of a helmet by taking into account key points of the head. Background Art

[0002] Standardized wearing of hard hats plays a vital role in the workplace, especially in construction sites, factories or other environments where there are risks of falling objects, collisions, etc. In many countries and regions, workplace safety regulations require employees to wear hard hats in specific environments. These regulations and standards have promoted the development and application of hard hat testing technology.

[0003] Traditional helmet wearing detection mainly relies on manual monitoring, which is inefficient and difficult to achieve comprehensive coverage of all workers. Manual inspections require a lot of time and energy, and are easily affected by subjective factors, such as fatigue and negligence of inspectors, which may lead to the omission of some cases where helmets are not worn correctly. Manual inspections cannot monitor the entire work area in real time, and spot checks are often carried out during the inspection period. This means that during the inspection interval, if a worker does not wear a helmet correctly, it is difficult to be discovered and corrected in time, thereby increasing the potential risk of safety accidents. In addition, in addition to manual labor, some traditional methods rely on specific technologies, such as sensor-based detection methods, which require the helmet to be modified, increasing cost and complexity. In addition, sensors may be affected by environmental factors such as vibration and temperature, leading to misjudgment.

[0004] There are also detection methods that use image recognition to identify the wearing of helmets. The camera is used to collect images and the image recognition technology is used to determine whether the person is wearing a helmet. This method is a non-contact detection method that does not require special requirements for personnel and can detect multiple people at the same time. However, this recognition method in the prior art can only determine whether a person is wearing a helmet, but cannot determine whether the wearing is standardized. If the helmet is not worn in a standardized manner, it will also cause great safety hazards. At the same time, relying solely on one type of image recognition will have a high probability of error. Therefore, there is an urgent need for a helmet wearing standard detection system that can quickly process video streams for instant feedback and warnings. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the above-mentioned prior art, and to provide a method and system for identifying the standard wearing of helmets considering the key points of the head. According to different angles and different ways of wearing helmets, a standard wearing helmet algorithm based on traditional visual algorithms combined with deep learning algorithms is proposed, especially target detection algorithms, key point detection and classification models. These technologies can automatically detect whether personnel wear helmets in accordance with safety regulations, which is of great significance for improving safety compliance in the workplace.

[0006] On the one hand, a method for identifying the proper wearing of a helmet by considering the key points of the head is provided, comprising the following steps: S1: Build and train a human body detection model and a standard helmet wearing classification model; S2: Collect surveillance video, extract frames from the video stream at specific time intervals to obtain video frames, use the trained human body detection model to locate the human body in the video frame and output a human body image; S3: Use the human key point recognition algorithm to infer each human image, locate the positions of several key points of the human body, take the relative positions between the key points of the head area to determine the head deflection posture, and crop the head area image according to the key points of the head area; S4: Inputting the head area image into the standard helmet wearing classification model for inference and recognition.

[0007] Furthermore, in step S1, the human body detection model is constructed based on the target detection algorithm, including: S101: Collect a large amount of human image data, mark the position of the human body in the image, construct a data set after preprocessing the data and its annotations, and split the data set into a training set and a validation set; S102: Select the target detection algorithm model, configure the model parameters, and define the loss function, including the position loss and confidence loss , constantly adjust the weights to minimize the loss function, where the position loss is Calculate the distance between the predicted box and the true box, expressed as follows: The confidence loss is calculated by the cross entropy loss model to calculate the difference between the confidence prediction of the target existence and the true label, which is expressed as follows: in, and are the predicted and true bounding box coordinates, respectively. is the smooth L1 loss function method, is the total number of predicted bounding boxes, Represents the index of the four coordinate parameters 1 to 4 of the bounding box, including the center point x coordinate, center point y coordinate, width and height, is the true label, is the confidence of the model prediction; S103: Use non-maximum suppression (NMS) to remove overlapping bounding boxes, sort the bounding boxes by confidence, select the box with the highest confidence, and calculate the intersection and union ratio of the box with other boxes. If the intersection-and-union ratio exceeds the set threshold, other boxes are removed and the intersection-and-union ratio is The calculation formula is as follows: in, and are two boxes to be intersected and compared; S104: Evaluate model performance on the validation set, and optimize the model according to the evaluation results.

[0008] Further, in step S1, the construction of the standard helmet wearing classification model includes: S111: collecting image data including wearing and not wearing a helmet, and performing corresponding category annotation to construct a data set, wherein the categories include standard wearing, irregular wearing, not wearing a helmet, unknown, and background, splitting the data set into a training set and a validation set, and extracting features that are helpful for classification from the image, including original pixel values, edge detection results, and color histograms; S112: Select a convolutional neural network to build a standard helmet wearing classification model, use the labeled training set to train the model, and define the loss function , the announcement is as follows: in, is the sample size, For the The true labels of samples, The model predicts The probability that a sample is a positive class; S113: Use gradient descent to minimize the loss function , the announcement is as follows: in, represents the model parameters, is the learning rate, Represents the loss function About parameters The gradient of S114: Comprehensively evaluate the performance of the standard-wearing helmet classification model on the validation set through accuracy, recall rate and F1 value, and adjust model parameters according to the evaluation results to improve model performance.

[0009] Further, in step S3, reasoning on each human image by a human key point recognition algorithm includes: S301: Loading a pre-trained human key point recognition model, pre-processing the human image to be input to meet the input requirements of the model, including adjusting the image size and normalizing pixels; S302: inputting the preprocessed human body image into the human body key point recognition model, outputting the coordinates and confidence of the human body key points, and extracting the key point information in the output; S303: Plotting the extracted key points on the original image for visualization and analysis.

[0010] Further, in step S3, taking the relative positions between the key points in the head region to determine the head deflection posture includes: S311: extract key points of the head from the detection result, including nose, left eye, right eye, left ear and right ear; S312: By calculating the midpoint between the two key points of the left ear and the right ear To draw a line between two key points, the calculation formula is as follows: in, and They are the coordinates of the key points of the left ear and the right ear respectively; S313: Calculate the relative position of the nose key point and the line connecting the two ear key points, and determine the deflection posture of the head relative to the front of the image according to the range of the relative position value. If the current head is judged to be a complete side profile, it is retained and the subsequent head area is cropped; Otherwise, discard the image frame and process the next frame.

[0011] Further, in step S3, clipping the head region image according to the key points of the head region includes: According to the minimum and maximum coordinates of the key points of the head, calculate the width and height of the minimum bounding rectangle covering all the key points of the head; To ensure that the head area is completely cropped, a certain enlargement ratio is set and the cropping area rectangle is calculated; The head area is cropped from the original image according to the calculated cropping area rectangle.

[0012] Furthermore, in step S4, inputting the head area image into the standard helmet wearing classification model for inference recognition further includes: Set the classification decision threshold for each category including standard wearing, irregular wearing, not wearing a helmet, unknown and background; The head area image input into the standard helmet wearing classification model outputs the probability distribution of the standard helmet wearing, compares the probability distribution with the class decision threshold of each category, and determines the category to which the input head area image belongs; Collect the classification output results of multiple consecutive frames, set a counter to continuously record the number of times classified as irregular wearing, and trigger an alarm when the set threshold is reached. If the classification output result is not irregular wearing or the alarm has been triggered, reset the counter.

[0013] On the other hand, a helmet wearing recognition system considering key points of the head is provided, comprising: Model building module, used to build and train human body detection model and standard helmet wearing classification model; An image acquisition module is used to acquire surveillance videos, extract frames from the video stream at specific time intervals to obtain video frames, and use a trained human body detection model to locate the human body in the video frame and output a human body image; The head positioning module is used to infer each human body image through the human body key point recognition algorithm, locate the positions of several key points of the human body, determine the head deflection posture by taking the relative positions between the key points of the head area, and crop the head area image according to the key points of the head area; The classification and reasoning module inputs the head area image into the standard wearing helmet classification model for reasoning and recognition.

[0014] In addition, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, any of the above-mentioned methods for identifying the standard wearing of a helmet taking into account the key point positions of the head is implemented.

[0015] At the same time, an electronic device is provided, including: one or more processors; a storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement any of the above-mentioned methods for identifying the standard wearing of a helmet taking into account the key point positions of the head.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention accurately identifies the human head by combining the target detection algorithm, the human key point recognition algorithm and the classification model, and more accurately determines whether the helmet is worn in a standard manner on the basis of detecting whether the helmet is worn or not; The present invention quickly processes the monitoring video stream, sets a timer to record the classification results of multiple consecutive frames, and provides real-time feedback and alarms to avoid single recognition errors. The present invention determines whether it is a complete profile by the position relationship of the key points of the head. When the profile is complete, the accuracy of judging whether the helmet is worn in a standard manner can be greatly improved. After filtering the key points, a classification model is used to determine whether the head area is wearing a helmet in a standard manner. In order to increase the stability of the system, the pre-trained target detection and key point model is used to reduce the probability of false detection, which greatly improves the stability and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of a method for identifying standard wearing of a helmet taking into account key points of the head according to the present invention; Figure 2 This is an overall flow chart combining various algorithms of the present invention; Figure 3 A schematic diagram of key points of a human body and a head according to the present invention; Figure 4 This is an example diagram of a head area image according to the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] The present invention proposes a standard helmet wearing algorithm based on traditional visual algorithms combined with deep learning algorithms for different angles and different ways of wearing helmets. Whether it is a complete side face is determined by the position relationship of key points on the head. When it is a complete side face, the accuracy of judging the standard helmet wearing can be greatly improved. After filtering the key points, a classification model is used to determine whether the head area is a standard helmet wearing area. In order to increase the stability of the system, pre-trained target detection and key point models are used to reduce the probability of false detection, which greatly improves the stability and reliability.

[0020] The specific implementation of the present invention is described below with reference to the accompanying drawings and embodiments.

[0021] like Figure 1As shown, this embodiment provides a method for identifying the standard wearing of a helmet considering the key points of the head, and the technical solution includes the following steps: S1: Build and train a human body detection model and a standard helmet wearing classification model; S2: Collect surveillance video, extract frames from the video stream at specific time intervals to obtain video frames, use the trained human body detection model to locate the human body in the video frame and output a human body image; S3: Use the human key point recognition algorithm to infer each human image, locate the positions of several key points of the human body, take the relative positions between the key points of the head area to determine the head deflection posture, and crop the head area image according to the key points of the head area; S4: Inputting the head area image into the standard helmet wearing classification model for inference and recognition.

[0022] Wherein, in step S1, constructing the human body detection model based on the target detection algorithm includes: Collect a large amount of human image data, mark the position of the human body in the image, construct a data set after preprocessing the data and its annotations, and split the data set into a training set and a validation set; Select the target detection algorithm model, configure the model parameters, and define the loss function, including the position loss and confidence loss , constantly adjust the weights to minimize the loss function, where the position loss is Calculate the distance between the predicted box and the true box, expressed as follows: The confidence loss is calculated by the cross entropy loss model to calculate the difference between the confidence prediction of the target existence and the true label, which is expressed as follows: in, and are the predicted and true bounding box coordinates, respectively. For the smooth L1 loss function method, is the total number of predicted bounding boxes, Represents the index of the four coordinate parameters 1 to 4 of the bounding box, including the center point x coordinate, center point y coordinate, width and height, is the true label, is the confidence of the model prediction; Use non-maximum suppression (NMS) to remove overlapping bounding boxes, sort the bounding boxes by confidence, select the box with the highest confidence, and calculate the intersection and union ratio of this box with other boxes. If the intersection-and-union ratio exceeds the set threshold, other boxes are removed and the intersection-and-union ratio is The calculation formula is as follows: in, and are two boxes to be intersected and compared; The model performance is evaluated on the validation set, and the model is optimized based on the evaluation results.

[0023] Secondly, the construction of the standard helmet wearing classification model further includes: Collect image data including wearing and not wearing a helmet, and perform corresponding category annotation to construct a data set, the categories include standard wearing, irregular wearing, not wearing a helmet, unknown and background, split the data set into a training set and a validation set, extract features that are helpful for classification from the image, including original pixel values, edge detection results and color histograms; Select convolutional neural network to standardize the construction of helmet wearing classification model, use the labeled training set to train the model, and define the loss function , the announcement is as follows: in, is the sample size, For the The true labels of samples, The model predicts The probability that a sample is a positive class; Use gradient descent to minimize the loss function , the announcement is as follows: in, represents the model parameters, is the learning rate, Represents the loss function About parameters The gradient of The performance of the standard-wearing helmet classification model is comprehensively evaluated on the validation set through accuracy, recall rate and F1 value, and the model parameters are adjusted according to the evaluation results to improve the model performance.

[0024] In step S3, each human image is inferred by a human key point recognition algorithm, including: Load the pre-trained human key point recognition model and pre-process the human image to be input to meet the input requirements of the model, including adjusting the image size and normalizing pixels; The preprocessed human image is input into the human key point recognition model, and the coordinates and confidences of 17 key points of the human body are output, including five key points of the head, namely, nose, left eye, right eye, left ear, and right ear. Figure 3 As shown, and extract the key point information in the output; Plot the extracted keypoints on the original image for easier visualization and analysis.

[0025] Then, the head posture is judged by the relative positions of the five key points of the head. The pictures with the head posture of a completely side face are retained, and the pictures that are not completely side faces are discarded.

[0026] Specifically, in this embodiment, key points of the head are first extracted from the detection result, including the nose, left eye, right eye, left ear and right ear; By calculating the midpoint between the two key points of the left ear and the right ear To draw a line between two key points, the calculation formula is as follows: in, and They are the coordinates of the key points of the left ear and the right ear respectively; Calculate the relative position of the nose key point and the line connecting the two ear key points, and determine the deflection posture of the head relative to the front of the image according to the range of the relative position value, such as Figure 4 As shown, If the current head is judged to be a complete side profile, it is retained (such as Figure 4 Right), and then crop the head area; Otherwise (such as Figure 4 Left) Discard this frame and process the next one.

[0027] Then we crop the head area image according to the key points of the head area, including: According to the minimum and maximum coordinates of the key points of the head, calculate the width and height of the minimum bounding rectangle covering all the key points of the head; To ensure that the head area is completely cropped, a certain enlargement ratio is set and the cropping area rectangle is calculated; The head area is cropped from the original image according to the calculated cropping area rectangle.

[0028] Finally, the human head area is identified through inference of the standard helmet wearing classification model to determine whether the straps are worn in a standard manner. If it is continuously judged that the helmet is worn in a standard manner, an alarm is triggered.

[0029] Specifically, in this embodiment, the classification decision thresholds for each category including standard wearing, irregular wearing, not wearing a helmet, unknown and background are set; The head area image input into the standard helmet wearing classification model outputs the probability distribution of the standard helmet wearing, compares the probability distribution with the class decision threshold of each category, and determines the category to which the input head area image belongs; Collect the classification output results of multiple consecutive frames, set a counter to continuously record the number of times classified as irregular wearing, and trigger an alarm when the set threshold is reached. If the classification output result is not irregular wearing or the alarm has been triggered, reset the counter.

[0030] The overall algorithm flow is as follows Figure 2 shown.

[0031] This embodiment also discloses a helmet standard wearing recognition system considering the key points of the head, including: Model building module, used to build and train human body detection model and standard helmet wearing classification model; An image acquisition module is used to acquire surveillance videos, extract frames from the video stream at specific time intervals to obtain video frames, and use a trained human body detection model to locate the human body in the video frame and output a human body image; The head positioning module is used to infer each human body image through the human body key point recognition algorithm, locate the positions of several key points of the human body, determine the head deflection posture by taking the relative positions between the key points of the head area, and crop the head area image according to the key points of the head area; The classification and reasoning module inputs the head area image into the standard wearing helmet classification model for reasoning and recognition.

[0032] Among them, the functional implementation of each module and unit in the system for identifying the standard wearing of a helmet considering the key points of the head corresponds to the steps in the above-mentioned embodiment of the method for identifying the standard wearing of a helmet considering the key points of the head, and its functions and implementation process will not be repeated here one by one.

[0033] Finally, it should be noted that the above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

[0034] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for identifying the proper wearing of a helmet by considering the key points of the head, characterized in that: The steps include: S1: Build and train a human body detection model and a standard helmet wearing classification model; S2: Collect surveillance video, extract frames from the video stream at specific time intervals to obtain video frames, use the trained human body detection model to locate the human body in the video frame and output a human body image; S3: Use the human key point recognition algorithm to infer each human image, locate the positions of several key points of the human body, take the relative positions between the key points of the head area to determine the head deflection posture, and crop the head area image according to the key points of the head area; S4: Inputting the head area image into the standard helmet wearing classification model for inference and recognition.

2. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 1, characterized in that: In step S1, constructing the human body detection model based on the target detection algorithm further includes: S101: Collect a large amount of human image data, mark the position of the human body in the image, construct a data set after preprocessing the data and its annotations, and split the data set into a training set and a validation set; S102: Select the target detection algorithm model, configure the model parameters, and define the loss function, including the position loss and confidence loss , constantly adjust the weights to minimize the loss function, where the position loss is Calculate the distance between the predicted box and the true box, expressed as follows: The confidence loss is calculated by the cross entropy loss model to calculate the difference between the confidence prediction of the target existence and the true label, which is expressed as follows: in, and are the predicted and true bounding box coordinates, respectively. is the smooth L1 loss function method, is the total number of predicted bounding boxes, Represents the index of the four coordinate parameters 1 to 4 of the bounding box, including the center point x coordinate, center point y coordinate, width and height, is the true label, is the confidence of the model prediction; S103: Use non-maximum suppression (NMS) to remove overlapping bounding boxes, sort the bounding boxes by confidence, select the box with the highest confidence, and calculate the intersection and union ratio of the box with other boxes. If the intersection-and-union ratio exceeds the set threshold, other boxes are removed and the intersection-and-union ratio is The calculation formula is as follows: in, and are two boxes to be intersected and compared; S104: Evaluate model performance on the validation set, and optimize the model according to the evaluation results.

3. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 1, characterized in that: In step S1, the construction of the standard helmet wearing classification model further includes: S111: collecting image data including wearing and not wearing a helmet, and performing corresponding category annotation to construct a data set, wherein the categories include standard wearing, irregular wearing, not wearing a helmet, unknown, and background, splitting the data set into a training set and a validation set, and extracting features that are helpful for classification from the image, including original pixel values, edge detection results, and color histograms; S112: Select a convolutional neural network to build a standard helmet wearing classification model, use the labeled training set to train the model, and define the loss function , the announcement is as follows: in, is the sample size, For the The true labels of samples, The model predicts The probability that a sample is a positive class; S113: Use gradient descent to minimize the loss function , the announcement is as follows: in, represents the model parameters, is the learning rate, Represents the loss function About parameters The gradient of S114: Comprehensively evaluate the performance of the standard-wearing helmet classification model on the validation set through accuracy, recall rate and F1 value, and adjust model parameters according to the evaluation results to improve model performance.

4. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 1, characterized in that: In step S3, reasoning on each human image by a human key point recognition algorithm further includes: S301: Loading a pre-trained human key point recognition model, pre-processing the human image to be input to meet the input requirements of the model, including adjusting the image size and normalizing pixels; S302: inputting the preprocessed human body image into the human body key point recognition model, outputting the coordinates and confidence of the human body key points, and extracting the key point information in the output; S303: Plotting the extracted key points on the original image for visualization and analysis.

5. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 4, characterized in that: In step S3, taking the relative positions between the key points in the head region to determine the head deflection posture further includes: S311: extract key points of the head from the detection result, including nose, left eye, right eye, left ear and right ear; S312: By calculating the midpoint between the two key points of the left ear and the right ear To draw a line between two key points, the calculation formula is as follows: in, and They are the coordinates of the key points of the left ear and the right ear respectively; S313: Calculate the relative position of the nose key point and the line connecting the two ear key points, and determine the deflection posture of the head relative to the front of the image according to the range of the relative position value. If the current head is judged to be a complete side profile, it is retained and the subsequent head area is cropped; Otherwise, discard the image frame and process the next frame.

6. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 5, characterized in that: In step S3, cutting the head region image according to the key points of the head region further includes: According to the minimum and maximum coordinates of the key points of the head, calculate the width and height of the minimum bounding rectangle covering all the key points of the head; To ensure that the head area is completely cropped, a certain enlargement ratio is set and the cropping area rectangle is calculated; The head area is cropped from the original image according to the calculated cropping area rectangle.

7. The method for identifying the proper wearing of a helmet by considering the key points of the head according to claim 1, characterized in that: In step S4, inputting the head area image into the standard helmet wearing classification model for reasoning and recognition further includes: Set the classification decision threshold for each category including standard wearing, irregular wearing, not wearing a helmet, unknown and background; The head area image input into the standard helmet wearing classification model outputs the probability distribution of the standard helmet wearing, compares the probability distribution with the class decision threshold of each category, and determines the category to which the input head area image belongs; Collect the classification output results of multiple consecutive frames, set a counter to continuously record the number of times classified as irregular wearing, and trigger an alarm when the set threshold is reached. If the classification output result is not irregular wearing or the alarm has been triggered, reset the counter.

8. A helmet wearing recognition system considering key head point positions, characterized in that: include: Model building module, used to build and train human body detection model and standard helmet wearing classification model; An image acquisition module is used to acquire surveillance videos, extract frames from the video stream at specific time intervals to obtain video frames, and use a trained human body detection model to locate the human body in the video frame and output a human body image; The head positioning module is used to infer each human image through the human key point recognition algorithm, locate the positions of several key points of the human body, determine the head deflection posture by taking the relative positions between the key points of the head area, and crop the head area image according to the key points of the head area; The classification and reasoning module inputs the head area image into the standard wearing helmet classification model for reasoning and recognition.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for identifying the standard wearing of a helmet taking into account the key point positions of the head as described in any one of claims 1-7 is implemented.

10. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the method for identifying the standard wearing of a helmet considering the key point positions of the head as described in any one of claims 1-7.

Images