A method for rapid identification of abnormal behaviors in children with autism spectrum disorder

Abstract

The present invention belongs to the technical field of behavior recognition, and provides a method for quickly recognizing abnormal behaviors of children with autism spectrum disorder. The method includes obtaining the position information and face orientations of all children in each frame of the video of children's free activities; analyzing the position movement and activity range of each child in different frame images based on the position information to obtain the activity tendency of each child; performing clustering analysis on the position information of all children in each frame image, and adjusting the clustering parameters in combination with the activity tendency to obtain the optimal comparison range of each child; analyzing the active and passive interaction situations between each child and other children within its corresponding optimal comparison range based on the face orientation to obtain the interaction tendency and interaction weight of each child; and combining the activity tendency, interaction tendency and interaction weight to obtain the behavior abnormality degree of each child, thereby completing the recognition of children's behavior abnormalities. The method provided by the present invention can quickly and accurately recognize children with behavior abnormalities based on the adjusted optimal comparison range.

Description

Technical Field

[0001] The present invention relates to the technical field of behavior recognition, and particularly relates to a method for quickly recognizing abnormal behaviors of children with autism spectrum disorder. Background Art

[0002] Autism spectrum disorder (ASD) is a neurodevelopmental disorder. Children with autism spectrum disorder may exhibit characteristics such as social withdrawal, delayed language development, strong interest in specific things, and repetitive behaviors. The recognition of abnormal behaviors of children with autism spectrum disorder monitors children's social interactions, communication abilities, and repetitive behaviors. By observing children's eye contact, language expression, play styles, and responses to environmental changes, combined with standardized scales and intelligent recognition technologies to screen potential autism characteristics, early recognition and intervention of children with autism spectrum disorder can significantly improve the quality of life and social abilities of children with autism spectrum disorder.

[0003] When using pose estimation and behavior recognition technologies to recognize abnormal behaviors of children with autism spectrum disorder, it is usually necessary to collect the behavior videos of children in their living environments for analysis. For example, at home or in kindergarten, the behavior videos of children in kindergarten can well show the performance of children when interacting and communicating with others and their behaviors when playing, and are suitable as analysis samples. When analyzing the interaction situation between children, if an inappropriate comparison object for analyzing the interaction is not selected, it will lead to inaccurate analysis of the children's interaction situation. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a method for quickly recognizing abnormal behaviors of children with autism spectrum disorder.

[0005] According to a method for quickly recognizing abnormal behaviors of children with autism spectrum disorder provided by the present invention, the method includes:

[0006] Obtain the position information and facial orientations of all children in each frame of an image of a video of children's free activities;

[0007] Based on the position information, analyze the position movement and activity range of each child in different frames of images to obtain the activity tendency of each child;

[0008] Perform clustering analysis on the position information of all children in each frame of image, and adjust the clustering parameters in combination with the activity tendency to obtain the best comparison range for each child;

[0009] Based on the facial orientations, analyze the active and passive interaction situations between each child and other children within its corresponding best comparison range to obtain the interaction tendency and interaction weight of each child;

[0010] Combining the activity tendency, the interaction tendency, and the interaction weight, obtain the behavior abnormality degree of each child;

[0011] Based on the behavior abnormality degree, identify children with behavior abnormalities.

[0012] In some embodiments of the present invention, obtaining the position information and face orientations of all children in each frame of an image of a free activity video of children includes:

[0013] Obtain a free activity video of children;

[0014] Using the YOLO model, identify the positions and face orientations of all children in each frame of the free activity video, and mark the outer contours of each child in the form of rectangular bounding boxes to obtain the position information and face orientations of all children in each frame of the image.

[0015] In some embodiments of the present invention, based on the position information, analyzing the position movement and activity range of each child in different frames of images to obtain the activity tendency of each child includes:

[0016] Based on the position information, analyze the position changes of each child in two adjacent frames of images to obtain a position movement parameter;

[0017] Based on the position information, analyze the maximum activity range of each child in all frames of images to obtain an activity range parameter;

[0018] According to the position movement parameter and the activity range parameter, obtain the activity tendency of each child.

[0019] In some embodiments of the present invention, the maximum activity range is the radius of the minimum circumscribed circle of all position information of a child in all frames of images.

[0020] In some embodiments of the present invention, perform clustering analysis on the position information of all children in each frame of the image, and adjust the clustering parameters in combination with the activity tendency to obtain the best comparison range of each child, including:

[0021] Analyze the Euclidean distance between the position information of any two children to obtain the initial similarity between any two children;

[0022] According to the activity tendency, identify active children;

[0023] Combining the activity tendency, adjust the initial similarity between the active children and other children to obtain the corrected similarity of the active children, and further obtain a corrected similarity matrix;

[0024] Based on the corrected similarity matrix, perform clustering analysis on the position information of all children in each frame of the image to obtain the clustering result of each frame of the image;

[0025] Analyze the clustering results of all frames of the image to obtain the optimal comparison range for each child.

[0026] In some embodiments of the present invention, analyzing the clustering results of all frames of the image to obtain the optimal comparison range for each child includes: among the clustering results of all frames of the image, taking the children who belong to the same cluster the most times and are the closest in distance to the th child as the optimal comparison range for the th child.

[0027] In some embodiments of the present invention, based on the facial orientation, analyze the active and passive interaction situations between each child and other children within its corresponding optimal comparison range to obtain the interaction tendency and interaction weight of each child, including:

[0028] Establish an orientation vector according to the facial orientation;

[0029] Denote the connection line between each child and other children within its corresponding optimal comparison range as the comparison reference line;

[0030] Based on the orientation vector and the comparison reference line, analyze the active and passive interaction situations between each child and other children within its corresponding optimal comparison range to obtain the interaction tendency and interaction weight of each child.

[0031] In some embodiments of the present invention, based on the orientation vector and the comparison reference line, analyze the active and passive interaction situations between each child and other children within its corresponding optimal comparison range to obtain the interaction tendency and interaction weight of each child, including:

[0032] Analyze the included angle relationship between the orientation vector of each child and the comparison reference line to obtain the first positional relationship;

[0033] Analyze the included angle relationship between the orientation vectors of other children within the corresponding optimal comparison range of each child and the comparison reference line to obtain the second positional relationship;

[0034] Combine the first positional relationship and the second positional relationship to obtain the interaction tendency of each child;

[0035] Obtain the interaction weight of each child according to the second positional relationship.

[0036] In some embodiments of the present invention, obtaining the behavior abnormality degree of each child by combining the activity tendency, the interaction tendency, and the interaction weight includes:

[0037] Taking the interaction weight as the weight of the interaction tendency, and taking 1 minus the interaction weight as the weight of the activity tendency, to obtain the behavior abnormality degree of each child.

[0038] In some embodiments of the present invention, identifying children with abnormal behaviors according to the behavior abnormality degree includes:

[0039] Presetting an abnormality threshold;

[0040] When the behavior abnormality degree of a child is greater than or equal to the abnormality threshold, it is identified that the child has abnormal behaviors;

[0041] Using the Childhood Autism Screening Scale and the Childhood Autism Rating Scale to evaluate the autism risk of the child.

[0042] As can be seen from the above embodiments, a method for quickly identifying abnormal behaviors of children with autism spectrum disorder provided by the embodiments of the present invention has the following beneficial effects:

[0043] The present invention obtains the position information and face orientations of all children in each frame of the free activity video of children, and is used to analyze the behavior performance of children in the activity state of independent actions; based on the position information, analyzes the position movement and activity range of each child in different frame images to obtain the activity tendency of each child. The activity tendency of children is an important indicator for judging children's abnormal behaviors, and to a certain extent, it can reflect the possibility that the child has autism spectrum disorder; performs clustering analysis on the position information of all children in each frame of image, and adjusts the clustering parameters in combination with the activity tendency to obtain the best comparison range for each child. The acquisition of the best comparison range provides a real and reliable comparison range for the subsequent analysis of children's interaction situations, and improves the accuracy of the final recognition result; based on the face orientation, analyzes the active and passive interaction situations between each child and other children within its corresponding best comparison range to obtain the interaction tendency and interaction weight of each child. The interaction tendency of children is another important indicator for judging children's abnormal behaviors, and to a certain extent, it can also reflect the possibility that the child has autism spectrum disorder; therefore, by combining the activity tendency, the interaction tendency, and the interaction weight, the behavior abnormality degree of each child is obtained; according to the behavior abnormality degree, children with abnormal behaviors are identified. The method provided by the present invention can quickly and accurately identify children with abnormal behaviors based on the adjusted best comparison range, in combination with the activity tendency and interaction tendency of children.

[0044] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present invention. Description of the Drawings

[0045] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0046] Figure 1 It is a schematic diagram of the basic process of a method for quickly identifying abnormal behaviors of children with autism spectrum disorder provided by an embodiment of the present invention. Detailed implementation manners

[0047] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the drawings and preferred embodiments, will detail the specific implementation manners, structures, features and effects of a method for quickly identifying abnormal behaviors of children with autism spectrum disorder proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. Such terms as "including", "comprising" or any other variant thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the article or device including the element.

[0049] The following will, in combination with the drawings, introduce in detail a method for quickly identifying abnormal behaviors of children with autism spectrum disorder provided in this embodiment.

[0050] Please refer to Figure 1 , which shows the basic process of a method for quickly identifying abnormal behaviors of children with autism spectrum disorder provided by an embodiment of the present invention.

[0051] As Figure 1 shown, a method for quickly identifying abnormal behaviors of children with autism spectrum disorder provided by an embodiment of the present invention specifically includes the following steps:

[0052] S100: Obtain the position information and facial orientations of all children in each frame of the video of children's free activities.

[0053] First, obtain the free-play videos of children in the kindergarten. Specifically, use the surveillance cameras in the kindergarten classrooms to obtain free-play videos with a length of 30 minutes (the time when children play and communicate independently without teacher guidance) at fixed positions. The video specification is 30 frames per second, which is used to analyze the behavioral performance of children in the state of independent activities.

[0054] Then, use the existing pre-trained YOLO model for person recognition to identify the positions and facial orientations of all children in each frame of the free-play video, and mark the outer contours of each child in the form of rectangular bounding boxes. The center of the rectangular bounding box represents the position of each child, and the position information and facial orientation of all children in each frame of the image are obtained.

[0055] S200: Based on the position information, analyze the position movement and activity range of each child in different frames of the image to obtain the activity tendency of each child.

[0056] Children with autism spectrum disorder tend to play alone during free play and are unable to actively integrate into interactive group activities. By analyzing the positions and movement trajectories of all children in the free-play videos of children, the activity tendency of each child can be obtained. The possible situations of children during free play are as follows: (1) Multiple children play together in groups. Such children have more interactions with the groups they are with, and will gather with other children in terms of position, but the overall movement trajectory may not change much; (2) Children play in multiple groups or alone. These are usually lively children with large changes in movements and movement trajectories, and may have more interactions with others; (3) Children play alone, such as introverted children or children with autism spectrum disorder, who are not good at actively integrating into the group and may play alone in the same position for a long time, hardly interacting with others.

[0057] First, analyze the overall movement of children. If a child has little position movement and a small activity range throughout the free-play video, it reflects that the child has a low activity tendency, and the child is more likely to be a child with autism spectrum disorder.

[0058] Based on the above analysis, in this embodiment, based on the position information, the position movement and activity range of each child in different frame images are analyzed to obtain the activity tendency of each child. Further, it includes analyzing the position change of each child in two adjacent frame images based on the position information, that is, analyzing the position change of each child in each frame image and its adjacent previous frame image, and recording the position change as the displacement of each child in each frame image, with the displacement of the first frame recorded as 0; calculating the cumulative sum of the displacements of the children in all frame images to obtain the position movement parameter. Based on the position information, the maximum activity range of each child in all frame images is analyzed simultaneously to obtain the activity range parameter, where the maximum activity range is the radius of the minimum circumscribed circle of all the position information of the child in all frame images. Then, according to the position movement parameter and the activity range parameter, the activity tendency of each child is obtained.

[0059] Construct the activity tendency of the th child as:

[0060]

[0061] In the formula, represents the activity tendency of the th child; represents the radius of the minimum circumscribed circle of all the positions of the th child in all frame images; represents the total number of frames of the free activity video; represents the displacement of the th child in the th frame image; represents the linear normalization function.

[0062] represents the cumulative sum of the displacement distances of the th child in all frame images, that is, represents the position movement parameter; reflects the activity range size of the th child in all frame images, that is, represents the maximum activity range; represents the degree of change in the movement trajectory of the th child. The larger this value is, the more active the th child is and the larger the activity range is, and the greater the activity tendency is.

[0063] Similarly, obtain the activity tendencies of all children.

[0064] S300: Perform clustering analysis on the position information of all children in each frame image, and adjust the clustering parameters in combination with the activity tendency to obtain the best comparison range for each child.

[0065] The activity tendency reflects the position activities of each child as a whole. There are also other situations, such as multiple children playing together or introverted children having a lower activity tendency. Just relying on the activity tendency of the overall position cannot completely and accurately show whether a child has a tendency to be lonely. Therefore, it is necessary to further analyze the local actions and interaction situations of each child.

[0066] Before analyzing the interaction situation of children, first cluster the distances between all children to analyze whether there are abnormalities in the actions of children and their communication and interaction with others through the correlation comparison between children in the same cluster. The closer the distance between children, the more likely they are playing together. Therefore, cluster the position information of children. If the comparison range is selected improperly, it may lead to incorrect judgment of the interaction object when judging the interaction situation of children later, resulting in errors in the judgment of the interaction situation of children.

[0067] Based on the above analysis, in this embodiment, cluster analysis is performed on the position information of all children in each frame of image, and the clustering parameters are adjusted in combination with the activity tendency to obtain the best comparison range for each child. Further, it includes analyzing the Euclidean distance between the position information of any two children to obtain the initial similarity between any two children; identifying active children according to the activity tendency; combining the activity tendency to adjust the initial similarity between active children and other children to obtain the corrected similarity of active children, and then obtaining the corrected similarity matrix; based on the corrected similarity matrix, performing cluster analysis on the position information of all children in each frame of image to obtain the clustering result of each frame of image; analyzing the clustering results of all frames of images to obtain the best comparison range for each child.

[0068] Taking any frame of image as an example, the specific elaboration is as follows: Use the AP clustering algorithm to cluster all children. Since there are children with a large activity tendency, these children may be far from the positions where other children gather in some frames of images during the activity process, resulting in the positions of their corresponding points showing certain outlier characteristics. However, this is actually due to their large activity tendency rather than being unsociable. The existence of such children may lead to inaccurate clustering results. Therefore, in order to obtain more accurate clustering results, use the activity tendency of children to adjust the similarity values of different individuals during clustering, making the clustering results more accurate, so as to obtain a more referenceable comparison range for each child. The specific operation steps are as follows:

[0069] Obtain the similarity and reference degree. One child corresponds to one point. Still use the center marked by the bounding box to represent the position of each child, and use the negative value of the Euclidean distance between every two points as the initial similarity between any two points and point between , is equivalent to. The similarity between every two points and the reference degree of each point to itself constitute a similarity matrix. For any point , the minimum value in the similarity matrix is taken as the reference degree of point . The similarity matrix and the reference degree are well-known technologies and will not be elaborated here. In order to better classify children with high activity tendencies into other clusters during clustering instead of showing outlier characteristics (too far from the cluster center), the initial similarity of children with high activity tendencies is adjusted. Therefore, an active threshold is preset, and the active threshold can be 0.7. All children with activity tendencies greater than or equal to the activity threshold are screened out and recorded as active children. Taking the th active child as an example, the initial similarity between the th active child and other children is adjusted as follows: th active child, the initial similarity between the th active child and other children is adjusted:

[0070]

[0071] In the formula, represents the corrected similarity between the th active child and another child; represents the initial similarity between the th active child and another child; represents any other child except the th active child; represents the activity tendency of the th active child; represents the active threshold.

[0072] represents the part by which the activity tendency of the th active child exceeds the active threshold. The value range of is [0.7, 1]. The greater the activity tendency , the smaller the value of . Since is a negative number (the negative of the Euclidean distance between the position point of the th active child and the position point of the th child), the greater the activity tendency, the greater the value of the adjusted similarity .

[0073] Similarly, the corrected similarities between all active children and all other children are obtained. The similarities of all children to all other children are combined to form a corrected similarity matrix , where the data corresponding to active children in the similarity matrix is the corrected similarity, and the data corresponding to other inactive children is the initial similarity.

[0074] Initialize the attraction matrix and the membership matrix . Usually, the attraction matrix and the membership matrix are initialized as zero matrices. Any element of the attraction matrix represents the attraction of point to point , that is, it represents the tendency of the th child within the cluster centered on it when the th child is used as a cluster center; any element of the membership matrix represents the availability of point as the clustering center of the th point. As the th point's clustering center availability.

[0075] Iteratively update the attraction matrix and the membership matrix . By modifying the similarity matrix and the membership matrix to perform the iterative update of the attraction matrix , and then through the attraction matrix to perform the iterative update of the membership matrix . This iterative update method is an existing technology and will not be elaborated here.

[0076] Set the decision matrix , where , . When is greater than 0, it is considered that point is a clustering center until the number of consecutive unchanged clustering centers obtained in the decision matrix exceeds the preset control parameter. Usually, the preset control parameter is taken as 15, that is, the clustering centers in the decision matrix do not change for 15 consecutive times. After the decision matrix converges, all points are assigned to the nearest clustering center to achieve clustering.

[0077] Similarly, obtain the clustering results of all frame images.

[0078] After obtaining the clustering results of each frame image, analyze the clustering results of all frame images to obtain the best comparison range for each child. Specifically, among the clustering results of all frame images, take the th child and the children who belong to the same cluster the most times and are the closest in distance. The value of The optimal comparison range of a child, and other children within the optimal comparison range are the comparison children corresponding to the child.

[0079] In this step, the AP clustering algorithm is used to cluster all children, and the similarity values in the clustering are adjusted in combination with the activity tendencies of each child to obtain the optimal comparison range for each child, so as to analyze the subsequent child interaction situation. Taking any child as an example, the other children included in the optimal comparison range of this child are the objects most likely to interact with this child. By analyzing the interaction situation between these objects and this child, it is possible to further determine whether this child has abnormal behavioral characteristics of autism spectrum disorder.

[0080] S400: Based on the face orientation, analyze the active and passive interaction situations between each child and other children within its corresponding optimal comparison range to obtain the interaction tendency and interaction weight of each child.

[0081] For introverted children and children with autism spectrum disorder, although both will show low activity in interactions, introverted children show normal interactions with others, while children with autism spectrum disorder can hardly complete normal interactions with others. For example, after being interacted with by others, they cannot make normal eye contact or verbal responses and hardly make eye contact with others. By analyzing the interaction situations between each child and other children in its optimal comparison range, the interaction tendency of each child can be obtained.

[0082] Based on the above analysis, in this embodiment, based on the face orientation, the active and passive interaction situations between each child and other children within its corresponding optimal comparison range are analyzed to obtain the interaction tendency and interaction weight of each child. Further, it includes: establishing a face orientation vector, where the starting point of the face orientation vector is the position of the child, and the magnitude of the face orientation vector is taken as a unit vector. Here, the face orientation vector is only used to represent the facing direction; the connection line between each child and any other child within its corresponding optimal comparison range is denoted as the comparison reference line, and there is a comparison reference line between any two children; based on the face orientation vector and the comparison reference line, the active and passive interaction situations between each child and other children within its corresponding optimal comparison range are analyzed to obtain the interaction tendency and interaction weight of each child. Furthermore, based on the face orientation vector and the comparison reference line, analyzing the active and passive interaction situations between each child and other children within its corresponding optimal comparison range to obtain the interaction tendency and interaction weight of each child includes: analyzing the included angle relationship between the face orientation vector of each child and the comparison reference line to obtain the first positional relationship; analyzing the included angle relationship between the face orientation vectors of other children within the optimal comparison range corresponding to each child and the comparison reference line to obtain the second positional relationship; combining the first positional relationship and the second positional relationship to obtain the interaction tendency of each child; and obtaining the interaction weight of each child according to the second positional relationship.

[0083] Construct the formula for calculating the interaction tendency of the

[0084]

[0085] In the formula, represents the interaction tendency of the th child; represents the total number of frames of the free activity video; represents the size of the optimal comparison range, where ; represents the th frame, and represents the included angle between the face orientation vector of the th child and the th comparison reference line, where the th comparison reference line is the connection line between the th child and its corresponding th comparison child; represents the th frame, and represents the included angle between the face orientation vector of the th comparison child corresponding to the th child and the th comparison reference line; Represents a linear normalization function.

[0086] Represents the cosine value of the angle between the face vector of the th child and the comparison baseline, that is, it represents the first positional relationship and reflects the initiative interaction situation of the th child with its th comparison child. The value range of this value is [-1, 1]. The larger this value is, the more consistent the face vector of the th child is with the direction of the comparison baseline, and the more likely the th child is to have an initiative interaction with its th comparison child; Represents the cosine value of the angle between the face vector of the th comparison child corresponding to the th child and the th comparison baseline, that is, it represents the second positional relationship and reflects the interaction situation of the th comparison child corresponding to the th child, that is, the passive interaction situation of the th child. When both of these values are larger, there is an interaction between the th child and its th comparison child, and the interaction tendency is greater. Since the face can only face one person, the maximum value is taken in the optimal comparison range Represents the cumulative sum of the interaction situations in all frame images. The larger this value is, the greater the interaction tendency of the

[0087] th child.

[0088] Similarly, obtain the interaction tendencies of all children. The greater the interaction tendency of a child, the more likely it is that the child has an effective interaction with other children. According to the second positional relationship, obtain the interaction weight of each child, that is, obtain the passive interaction situation of the th child in all frame images and analyze the interaction weight. The calculation formula for constructing the interaction weight of the

[0089]

[0090] In the formula, Represents the interaction weight of the th child; Represents the total number of frames of the free activity video; Indicates the size of the optimal comparison range, where ; Indicates the angle between the orientation vector of the th child in the th frame and the th comparison baseline for the th comparison child; Indicates the cosine function; Indicates the maximum function. That is, the average value of the passive interaction of the

[0091] S500: Combine the activity tendency, interaction tendency, and interaction weight to obtain the behavior abnormality degree of each child.

[0092] Combine the activity tendency, interaction tendency, and interaction weight to obtain the behavior abnormality degree of each child, including: using the interaction weight as the weight of the interaction tendency, and using 1 minus the interaction weight as the weight of the activity tendency to obtain the behavior abnormality degree of each child.

[0093] Construct the calculation method for the behavior abnormality of the th child as follows: In the formula,

[0094]

[0095] where indicates the behavior abnormality degree of the th child; indicates the interaction weight of the th child; indicates the activity tendency of the th child; indicates the interaction tendency of the th child; indicates the exponential function with the natural constant as the base; indicates the inverse proportional normalization function, where x is the independent variable.

[0096] Indicates that when the interaction weight is small, the th child has less interaction with other children, and the activity tendency is the main reference for calculating the behavior abnormality degree; Indicates that when the interaction weight is large, the interaction tendency is the main reference for calculating the behavior abnormality degree.

[0097] Similarly, obtain the behavior abnormality degrees of all children.

[0098] S600: Identify children with behavior abnormalities based on the behavior abnormality degrees.

[0099] Identify children with abnormal behaviors according to the degree of behavioral abnormality. Specifically, preset an abnormal threshold, and the value of the abnormal threshold can be 0.7; when the degree of behavioral abnormality of a child is greater than or equal to the abnormal threshold, identify that the child has abnormal behaviors; through guardians and teachers, use standardized assessment scales such as M-CHAT (Modified Checklist for Autism in Toddlers) and CARS (Childhood Autism Rating Scale) to evaluate the autism risk of children. Intervene in the abnormal behaviors of children with autism spectrum disorder in a timely manner according to the evaluation results and seek the help of professional physicians.

[0100] It should be noted that: the above-mentioned order of the embodiments of the present invention is only for description and does not represent the advantages or disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0101] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for quickly identifying abnormal behaviors of children with autism spectrum disorders, characterized in that: The method comprises: Obtain the position information and facial orientation of all children in each frame of the video of children's free activities; Based on the position information, analyzing the position movement and activity range of each child in different frame images to obtain the activity tendency of each child; Performing cluster analysis on the position information of all children in each frame of the image, and adjusting the clustering parameters in combination with the activity tendency to obtain the best contrast range for each child; Based on the facial orientation, analyzing active and passive interactions between each child and other children within the corresponding optimal comparison range to obtain an interaction tendency and an interaction weight of each child; Combining the activity tendency, the interaction tendency and the interaction weight, obtaining a behavior abnormality degree of each child; identifying children with behavioral abnormalities based on the degree of behavioral abnormality; Among them, the method for obtaining the optimal comparison range is: analyzing the Euclidean distance between the position information of any two children to obtain the initial similarity between any two children; identifying the active child according to the activity tendency; adjusting the initial similarity between the active child and other children in combination with the activity tendency to obtain the corrected similarity of the active child, and then obtaining a corrected similarity matrix, wherein the data corresponding to the active child in the corrected similarity matrix is ​​the corrected similarity, and the data corresponding to other inactive children is the initial similarity; based on the corrected similarity matrix, clustering analysis is performed on the position information of all children in each frame image to obtain the clustering result of each frame image; among the clustering results of all frame images, the child with the same height as the first frame image is selected. The children belong to the same cluster the most times and are closest to each other children, as the first The optimal contrast range for each child.

2. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 1, characterized in that: Get the location information and facial orientation of all children in each frame of the video of children's free activities, including: Get videos of children doing free activities; The YOLO model is used to identify the positions and facial orientations of all children in each frame of the free activity video, and the outer contour of each child is marked in the form of a rectangular bounding box to obtain the position information and facial orientation of all children in each frame.

3. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 1 or 2, characterized in that: Based on the position information, the position movement and activity range of each child in different frame images are analyzed to obtain the activity tendency of each child, including: Based on the position information, analyzing the position change of each child in two adjacent frames of images to obtain a position movement parameter; Based on the position information, analyzing the maximum activity range of each child in all frame images to obtain activity range parameters; The activity tendency of each child is obtained according to the position movement parameter and the activity range parameter.

4. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 3, characterized in that: The maximum activity range is the minimum circumscribed circle radius of all position information of the child in all frame images.

5. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 1, characterized in that: Based on the facial orientation, active and passive interactions between each child and other children within the corresponding optimal contrast range are analyzed to obtain the interaction tendency and interaction weight of each child, including: establishing a facing vector according to the facial orientation; Recording the line connecting each child and other children in the corresponding optimal comparison range as a comparison baseline; Based on the orientation vector and the comparison baseline, the active and passive interactions between each child and other children within the corresponding optimal comparison range are analyzed to obtain the interaction tendency and interaction weight of each child.

6. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 5, characterized in that: Based on the orientation vector and the comparison baseline, the active and passive interactions between each child and other children in the corresponding optimal comparison range are analyzed to obtain the interaction tendency and interaction weight of each child, including: Analyzing the angle relationship between the orientation vector of each child and the comparison baseline to obtain a first position relationship; Analyze the angle relationship between the orientation vectors of other children within the optimal comparison range corresponding to each child and the comparison baseline to obtain a second position relationship; Combining the first position relationship and the second position relationship to obtain the interaction tendency of each child; According to the second position relationship, the interaction weight of each child is obtained.

7. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 1, characterized in that: Combining the activity tendency, the interaction tendency and the interaction weight, the behavior abnormality degree of each child is obtained, including: The interaction weight is used as the weight of the interaction tendency, and 1 minus the interaction weight is used as the weight of the activity tendency to obtain the behavior abnormality degree of each child.

8. The method for rapidly identifying abnormal behaviors of children with autism spectrum disorders according to claim 1 or 7, characterized in that: Based on the degree of behavioral abnormality, children with behavioral abnormalities are identified, including: Preset abnormal thresholds; When the degree of abnormal behavior of the child is greater than or equal to the abnormal threshold, identifying that the child has abnormal behavior; The autism risk of the children was assessed using the Child Autism Screening Scale and the Child Autism Assessment Scale.

Images