CN115690043A - Queuing statistical method, device, equipment and medium based on multi-part detection - Google Patents
Queuing statistical method, device, equipment and medium based on multi-part detection Download PDFInfo
- Publication number
- CN115690043A CN115690043A CN202211343193.1A CN202211343193A CN115690043A CN 115690043 A CN115690043 A CN 115690043A CN 202211343193 A CN202211343193 A CN 202211343193A CN 115690043 A CN115690043 A CN 115690043A
- Authority
- CN
- China
- Prior art keywords
- list
- elements
- queuing
- sets
- intersection
- 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.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a queuing statistical method, a device, equipment and a medium based on multi-part detection, which are used for preprocessing by acquiring picture data; detecting the head, the upper half body, the lower half body and the whole body of each person in the preprocessed picture data by using a detection model, and acquiring a set; according to the intersection condition between every parts, every set is recombined and merged respectively to obtain the sets respectively corresponding to the individuals; calculating a vertex, a central point, a bottom point and inclination according to the coordinates of the set elements; and arranging according to the sets to obtain a list T, performing intersection judgment, screening and sequencing on elements in the list T, and performing approximate straight line judgment to determine a queuing list and the corresponding queuing number. The invention can realize the queuing counting function by only using one 2D image, can detect the parts of the head, the upper half body, the lower half body, the whole body and the like which are only partially visible when people are mutually shielded in a dense scene, and can improve the detection and the accuracy under shielding.
Description
Technical Field
The invention relates to the technical field of image detection and image tracking. In particular to a queuing statistical method, a device, equipment and a medium based on multi-part detection.
Background
With the application of machine learning in computer vision, particularly a deep neural network algorithm based on a CNN convolutional network, a human pedestrian detection algorithm in a general scene is relatively mature. Common algorithms such as YOLO series, SSD series, fasterrnnn, etc. are successfully applied in pedestrian detection. The common crowd counting or crowd density estimation method based on deep learning is greatly improved, and the crowd counting method for market scenic spots and the like saves manpower and reduces cost. However, the above population counting methods are all used to evaluate the number of all people appearing in a scene, and are difficult to apply when the population counting methods are used for specially subdividing application scenes, such as screening the number of people queued in a picture. In addition, the crowd estimation method is based on the overall characteristics of people and used for judging whether the people are in the queue or not, and the people cannot be accurately judged.
It is an important requirement that there be people in line and the count of the number of people in line within the scene. For example, when buying goods in a queue in a shopping mall, removing the queue count of an irrelevant person when entering or exiting various places such as scenic spots and the like is a very direct requirement. Thus, a method is continued that can detect and count the number of people in line.
Disclosure of Invention
In view of the above-mentioned shortcomings of the prior art, it is an object of the present invention to provide a queuing statistical method, apparatus, device and medium based on multi-part detection, which are used to solve the problem of people counting in the prior art.
To achieve the above and other related objects, the present invention provides a queuing statistics method based on multi-location detection, the method comprising: step S1: acquiring and preprocessing picture data;
step S2: detecting the head, the upper half body, the lower half body and the whole body of each person in the preprocessed picture data by using a detection model, representing each part by adopting a preset data structure containing coordinate information, and acquiring sets A, B, C and D respectively corresponding to the head, the upper half body, the lower half body and the whole body; and step S3: respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P; and step S4: respectively calculating a vertex, a central point, a bottom point and an inclination corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P; step S5: performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R; step S6: and (4) carrying out approximate straight line judgment on each element in the list R so as to determine the queuing list and the corresponding queuing number.
In an embodiment of the present invention, the acquiring and preprocessing the picture data includes: resetting the picture size of the acquired picture data to [3, M, N ]; wherein 3 is the number of channels after reset, M is the height after reset, and N is the width after reset; the picture data are 3 matrixes with the size of M × N, and matrix values are pixel values of the picture data; and subtracting the respective mean value of the 3 matrixes from the pixel value, and dividing the mean value by the respective variance of the 3 matrixes to obtain the normalized picture data.
In an embodiment of the invention, the obtaining of the detection model includes: marking the head, the upper half body, the lower half body and the whole body of all people in the picture respectively for each piece of picture data after preprocessing; dividing the marked picture data into a training set and a test set according to a certain proportion; respectively selecting a plurality of target detection algorithms, and carrying out iterative training on the picture data and the marked target part in the training set until the loss function of the training set is not reduced and the average precision mean value of the test set is not improved at the same time, so as to obtain a training model corresponding to each target detection algorithm; and testing each training model on the test set, and selecting the training with the average precision value as a detection model.
In an embodiment of the present invention, the preset data structure is: { class, [ x1, y1, x2, y2] }; wherein class represents a target category, takes values of 1-4 and respectively corresponds to the head, the upper body, the lower body and the whole body; [ x1, y1, x2, y2] represents the target coordinates, x1 is the abscissa of the upper left corner, y1 is the ordinate of the upper left corner, x2 is the abscissa of the lower right corner, and y2 is the abscissa of the lower right corner.
In an embodiment of the invention, the method of step S3 includes: step S301: respectively creating sets Pi, pk, pm and Pn to represent individuals corresponding to elements in the sets A, B, C and D, and putting the individuals into a total set P; step S302: according to the sequence from the set D to the set A, sequentially searching for intersection part elements which satisfy a preset intersection area ratio with each corresponding individual element in the set in all the previous item sets; step S303: moving the intersecting position elements in the previous item set to the set corresponding to the corresponding individual, and deleting the intersecting position elements from the previous item set; step S304: and circulating the steps to obtain sets Pi, pk, pm and Pn respectively corresponding to the individuals.
In an embodiment of the present invention, the method for obtaining the set Pi in the step S3 includes: step S311: creating a set Pi to represent individuals corresponding to each element di in the set D, adding each element di into the set Pi, and putting the set Pi into a total set P; step S312: sequentially calculating the intersection area of each element cj and di in the set C and dividing the intersection area by the area of cj to obtain an area ratio sj; step S313: if the area ratio sj is larger than a preset value Sc, the cj and the di belong to the same person, the cj is added into the set Pi, and the cj is deleted from the set C; step S314: according to the steps S312 and S313, the intersection condition of each element bj and di in the set B is sequentially calculated, bj meeting the preset intersection area ratio is added into the set Pi, and bj is deleted from the set B; step S315: according to the steps S312 and S313, the intersection condition of each element aj and di in the set A is sequentially calculated, aj meeting the preset intersection area ratio is added into the set Pi, and aj is deleted from the set A.
In an embodiment of the present invention, the method of step S313 further includes: if the set Pi contains the elements cj in the sets C after the round-robin of the set C, selecting cj with the largest area ratio of intersection with di, deleting the rest cj from the set Pi and adding the cj back to the set C; and/or if a plurality of elements cj with the same intersection area ratio exist in the set Pi after the set C is circularly circulated, calculating the central point distance between cj and di, keeping the cj corresponding to the minimum central point distance, and deleting the rest cj from the set Pi and adding the cj back to the set C.
In an embodiment of the present invention, the center point distance between cj and di is calculated as follows: d is a radical of center ＝sqrt{[(x1 cj +x2 cj )/2-(x1 di +x2 di )/2]*2+[(y1 cj +y2 cj )/2-(y1 di +y2 di )/2]*2}; wherein x1, x2, y1 and y2 respectively represent coordinates corresponding to the elements cj and di; sqrt represents the square root computation.
In an embodiment of the present invention, the performing intersection judgment, screening, and sorting on each element in the list T to obtain the list R includes: creating a set Qi representing a set of all elements intersected by the element ti in the list T, and adding the ti into the set Qi; sequentially judging whether the sets Q1 to Qi-1 already contain ti, if a certain set Qk contains ti, deleting the set Qi, and designating Qk as the set to which ti belongs; sequentially calculating the intersection ratio of the elements tj and ti in the list T, and if the intersection ratio is greater than a preset value S iou If so, the tj is considered to be intersected with the ti, and the tj is added into the set Qi or the set Qk; sequentially judging elements ti in the list T, if the number of the corresponding sets Qi is smaller than a preset threshold number Sq, determining that the ti is not in one queue, and deleting the ti and the sets Qi from the list T; and sorting the elements qj in the set Qi intersected with ti for any element ti in the list T, and sorting the elements qj in the set Qi in an ascending or descending manner according to the central point distance between qj and ai to obtain a sorted list R.
In an embodiment of the invention, the step S6 includes: carrying out neighborhood sorting on each element in the list R to judge any one or more of inclination, a central point straight line, a vertex straight line and a bottom point straight line; if the list R has a preset threshold value Sn which meets the judgment condition of at least one of the inclination, the straight line of the central point, the straight line of the top point and the straight line of the bottom point, the list R is judged to be a queuing queue, and the number of the elements in the list R is the number of queuing people.
To achieve the above and other related objects, the present invention provides a queuing statistics apparatus based on multi-part detection, the apparatus comprising: the acquisition module is used for acquiring and preprocessing the picture data; a processing module, configured to detect the head, the upper body, the lower body, and the whole body of each person in the preprocessed picture data by using a detection model, represent each part by using a preset data structure including coordinate information, and acquire sets a, B, C, and D corresponding to the head, the upper body, the lower body, and the whole body, respectively; respectively recombining and merging the sets A, B, C and D according to the intersection condition among all parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P; respectively calculating a vertex, a central point, a bottom point and an inclination corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P; performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R; and (4) carrying out approximate straight line judgment on each element in the list R so as to determine the queuing list and the corresponding queuing number.
To achieve the above and other related objects, the present invention provides a computer apparatus, comprising: a memory and a processor; the memory stores a computer program that the processor executes to implement the method as described above.
To achieve the above and other related objects, the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method as described above.
As described above, the queuing statistical method, the queuing statistical device, the queuing statistical equipment and the queuing statistical medium based on multi-position detection are characterized in that picture data are obtained and preprocessed; detecting the head, the upper half body, the lower half body and the whole body of each person in the preprocessed picture data by using a detection model, representing each part by adopting a preset data structure containing coordinate information, and acquiring sets A, B, C and D respectively corresponding to the head, the upper half body, the lower half body and the whole body; respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P; respectively calculating a vertex, a central point, a bottom point and an inclination corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P; performing ascending or descending order according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sorting on elements in the list T to obtain a list R; and (4) performing approximate straight line judgment on each element in the list R to determine the queuing list and the corresponding queuing number.
Has the following beneficial effects:
the invention can realize a relatively accurate queuing counting function by only using one 2D image, can detect a plurality of parts such as the head, the upper half body, the lower half body, the whole body and the like which are only partially visible when people are shielded by a person under a dense scene, can improve the detection and accuracy rate under shielding, and further saves the manpower and cost for queuing statistics.
Drawings
FIG. 1 is a flow chart illustrating a queuing statistics method based on multi-site detection according to an embodiment of the present invention.
Fig. 2 is a flowchart illustrating step S3 according to an embodiment of the present invention.
Fig. 3 is a flowchart illustrating the process of obtaining the set Pi in step S3 according to an embodiment of the present invention.
FIG. 4 is a block diagram of a multi-bit detection-based queue statistics apparatus according to an embodiment of the present invention.
FIG. 5 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
The following embodiments of the present invention are provided by way of specific examples, and other advantages and effects of the present invention will be readily apparent to those skilled in the art from the disclosure herein. The invention is capable of other and different embodiments and of being practiced or of being carried out in various ways, and its several details are capable of modification in various respects, all without departing from the spirit and scope of the present invention. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that the drawings provided in the following embodiments are only for illustrating the basic idea of the present invention, and the drawings only show the components related to the present invention rather than being drawn according to the number, shape and size of the components in actual implementation, and the type, amount and proportion of each component in actual implementation can be changed freely, and the layout of the components can be more complicated.
The existing crowd counting method is mainly used for counting the number of all people in a scene, but cannot be applied to subdividing the scene. For example, counting the number of people in a queue requires to eliminate people who are not in the queue, and the existing people counting method cannot solve the problem.
In order to solve the problems, the invention provides a queuing statistical method based on multi-part detection, which can realize a relatively accurate queuing counting function by only using one 2D image, can detect parts of a head, an upper half body, a lower half body, a whole body and the like which are only partially visible when people are shielded mutually in a dense scene, and can improve the detection and accuracy rate under shielding so as to further save the labor and the cost of queuing statistics.
Fig. 1 is a flow chart of a queuing statistics method based on multi-location detection according to an embodiment of the present invention. As shown, the method comprises:
step S1: and acquiring and preprocessing picture data.
In an embodiment of the present application, the acquiring and preprocessing the picture data includes:
step S101: resetting the picture size of the acquired picture data to [3, M, N ]; wherein 3 is the number of channels after reset, M is the height after reset, and N is the width after reset; the picture data are 3 matrixes with the size of M × N, and matrix values are pixel values of the picture data;
step S102: and subtracting the respective mean value of the 3 matrixes from the pixel value, and dividing the mean value by the respective variance of the 3 matrixes to obtain the normalized picture data.
For example, for the input picture data, the input shape is (C, H, W) (C represents the number of channels, H represents height, and W represents width), the size of the picture data is first reset to (3, m, n), for example, reset to (3, 512), where 3 represents the number of channels after reset, 512 represents height after adjustment, and 512 represents width after reset, and at this time, the picture data becomes 3 matrices of 512 × 512 size, the matrix value is the pixel value of the picture, the pixel value subtracts the mean value (123.675, 116.28, 103.53) of each of the 3 matrices, and then the variance (58.395, 57.12, 57.375) of each of the 3 matrices is divided to obtain the normalized picture data, and the size is still (3, 512).
Step S2: the method includes the steps of detecting the head, the upper body, the lower body and the whole body of each person in the preprocessed picture data by using a detection model, representing each part by adopting a preset data structure containing coordinate information, and acquiring sets A, B, C and D corresponding to the head, the upper body, the lower body and the whole body respectively.
In an embodiment of the present application, the method for acquiring and training the detection model includes:
step S201: marking the head, the upper half body, the lower half body and the whole body of all people in the picture respectively for each piece of picture data after preprocessing;
step S202: dividing the marked picture data into a training set and a test set according to a certain proportion;
step S203: respectively selecting a plurality of target detection algorithms, and carrying out iterative training on the picture data and the marked target part in the training set until the loss function of the training set is not reduced and the average precision mean value of the test set is not improved so as to obtain a training model corresponding to each target detection algorithm;
step S204: and testing each training model on the test set, and selecting the training with the average precision mean value to be stored as the detection model.
In one or more embodiments, the present invention provides an algorithm for judging queuing and counting the number of people in the queue based on a general target detection method, wherein the method for obtaining and training the detection model can be exemplified as follows:
step one, labeling pictures: for a piece of picture data, the coordinates of the head, upper body, lower body, and whole body of all the persons in the picture are marked respectively. The coordinates consist of the upper left corner and the lower right corner of the target (x 1, y1, x2, y 2), x1 is the upper left corner abscissa, y1 is the upper left corner ordinate, x2 is the lower right corner abscissa, and y2 is the lower right corner abscissa.
Step two, data division: assuming that the total number of data is L, the data can be divided into training sets and test sets according to a certain ratio, such as 8.
Step three, training: selecting an object detection algorithm, the object detection algorithm comprising: any one or more of yolov4, yolov5, yolov7, cascade _ rcnn, as the present invention uses yolov4 in some embodiments, iteratively trains the pictorial data and labeled targets in the training set until the training set loss function no longer falls and the test set mean of accuracy (mAP) no longer rises, to arrive at a training model for each of the target detection algorithms. Preferably, the model weights are saved every 10 iterations during the training process.
Step four, selecting a model: and testing the training model stored in the training process on the test set, and selecting the model with the highest mAP as the BEST model, namely as the detection model.
In an embodiment of the present application, the preset data structure is:
{class,[x1,y1,x2,y2]}；
wherein class represents a target category, takes values of 1-4 and respectively corresponds to the head, the upper body, the lower body and the whole body; [ x1, y1, x2, y2] represents the target coordinates, x1 is the upper left abscissa, y1 is the upper left ordinate, x2 is the lower right abscissa, and y2 is the lower right abscissa.
In brief, each part can be represented by the data structure, which contains the category of the corresponding part and can be used for dividing the set of different parts; and coordinate information is also included, for example, x1 is an upper left-corner abscissa, y1 is an upper left-corner ordinate, x2 is a lower right-corner abscissa, and y2 is a lower right-corner abscissa, and the coordinate information can be used for calculating subsequent vertexes, central points, bottom points, gradients and the like.
In addition, the invention also divides the detected different parts into the sets of corresponding parts, such as all head targets as a set { A }, all upper body targets as a set { B }, all lower body targets as a set { C }, and all whole body targets as a set { D }.
And step S3: and respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to the individuals, and putting all the sets into a total set P.
In an embodiment of the present application, the method of step S3 includes:
step S301: respectively creating sets Pi, pk, pm and Pn to represent individuals corresponding to elements in the sets A, B, C and D, and putting the individuals into a total set P;
step S302: sequentially searching for intersecting part elements which satisfy a preset intersecting area ratio with each corresponding individual element in the current item set in all the previous item sets according to the sequence from the set D to the set A;
step S303: moving the intersecting position elements in the previous item set to the set corresponding to the corresponding individual, and deleting the intersecting position elements from the previous item set;
step S304: and circulating the steps to obtain sets Pi, pk, pm and Pn respectively corresponding to the individuals.
In brief, the set D sequentially judges the intersecting elements in the previous item sets C, B, and a, moves the intersecting elements to D, then the set C sequentially judges the intersecting elements in the previous item sets B and a, moves the intersecting elements to C, then the set B sequentially judges the intersecting elements in the previous item set a, moves the intersecting elements to B, and finally forms new sets Pi, pk, pm, pn corresponding to individuals.
Specifically, the method for obtaining the set Pi in step S3 includes:
step S311: respectively creating a set Pi to represent the person corresponding to each element di in the set D, adding each element di into the set Pi, and putting the set Pi into a total set P;
step S312: sequentially calculating the intersection area of each element cj and di in the set C and dividing the intersection area by the area of cj to obtain an area ratio sj;
step S313: if the area ratio sj is larger than a preset value Sc, the cj and the di belong to the same person, the cj is added into the set Pi, and the cj is deleted from the set C;
in an embodiment of the present application, the method of step S313 further includes:
A. if the set Pi contains a plurality of elements cj in the set C after the round-robin of the set C, selecting cj with the largest area ratio of intersection with di, deleting the rest cj from the set Pi and adding the cj back to the set C;
and/or the presence of a gas in the atmosphere,
B. if a plurality of elements cj with the same intersection area ratio exist in the set Pi after the set C is circularly circulated, the central point distance between cj and di is calculated, the cj corresponding to the minimum central point distance is reserved, and the rest cj are deleted from the set Pi and added back to the set C.
Step S314: according to the steps S312 and S313, the intersection condition of each element bj and di in the set B is sequentially calculated, bj meeting the preset intersection area ratio is added into the set Pi, and bj is deleted from the set B;
step S315: according to the steps S312 and S313, the intersection condition of each element aj and di in the set A is sequentially calculated, aj meeting the preset intersection area ratio is added into the set Pi, and aj is deleted from the set A.
In some embodiments, the operations performed on the elements di within the set { D } in turn may be illustrated as follows:
first, a set { Pi } is created representing the person for di, di is added to { Pi }, and { Pi } is added to { P }.
Then, calculating the area of the intersection of the element cj and di in the set { C } in turn, and dividing the area by the area of cj to obtain the area ratio sj:
sj＝area(cj∩di)/area(cj)；
if sj is larger than a preset threshold Sc, then cj and di are considered to belong to the same person, cj is added into { Pi }, and cj is deleted from the set { C }.
If the { Pi } contains a plurality of elements cj in the { C } after the round-robin of the set { C }, firstly selecting cj corresponding to the maximum value of the di area ratio pj, and deleting the rest C from the { Pi } and adding the C into the { C }; and/or if a plurality of same pj values exist, calculating the distance d _ center between the cj center point and the di center point, wherein the calculation formula is as follows:
d center ＝sqrt{[(x1 cj +x2 cj )/2-(x1 di +x2 di )/2]*2+[(y1 cj +y2 cj )/2-(y1 di +y2 di )/2]*2}；
wherein x1, x2, y1 and y2 respectively represent coordinates corresponding to the elements cj and di; sqrt represents the square-on computation. Then, d is reserved center The cj corresponding to the minimum value, and the rest C are deleted from { Pi } and added to { C }.
Then, the relation between the bj and di in the set { B } is determined by the same method as steps S312 and S313, and the bj meeting the condition is added to { Pi }.
Finally, the relation between the element aj and the element di in the set { A } is judged by the same method as the steps S312 and S313, and the { Pi } is added to the aj meeting the condition.
In some embodiments, the operations performed on the elements di within the set { C } in turn may be exemplified as follows:
a set { Pk } is created representing the person to which ck corresponds, adding ck to { Pk } and adding { Pk } to { P }.
The relationship between the element bj and ck in the set { B } is judged by the same method as steps S312 and S313, and the bj meeting the condition is added into { Pk }.
And (4) judging the relationship between the element aj and ck in the set { A } by using the same method as the steps S312 and S313, and adding the eligible aj into { Pk }.
In some embodiments, the operations performed on the elements di within the set { B } in turn may be illustrated as follows:
and creating a set { Pm } to represent a person corresponding to bm, adding bm into { Pm } and adding { Pm } into { P }.
And (4) judging the relationship between the element aj and the bm in the set { A } by using the same method as the steps S312 and S313, and adding { Pm } into aj meeting the condition.
In some embodiments, the operations performed on the elements di within the set { A } in turn may be exemplified as follows:
a set { Pn } is created representing the person to which an corresponds, an is added to { Pn }, and { Pn } is added to { P }.
And step S4: and respectively calculating the vertex, the central point, the bottom point and the inclination corresponding to each set Pi according to the coordinates of the elements in each set Pi in the total set P.
It should be noted that the coordinates according to which the top point, the center point, the bottom point, and the gradient are calculated are the coordinate data detected by the monitoring model in step S2, and are represented by a preset data structure { class, [ x1, y1, x2, y2] }.
In an embodiment of the present application, the method for calculating the vertex in step S4 includes:
if set Pi includes element ai in set a, let ai be coordinates [ x1, y1, x2, y2], then the formula for calculating vertex vi of set Pi is: vi = [ (x 1+ x 2)/2, y1];
if set Pi does not contain element ai in set A, vertex vi of set Pi is empty.
In an embodiment of the present application, the method for calculating the center point in the step S4 includes:
if the set Pi includes the element bi in the set B, the element ci in the set C, and the element di in the set D, respectively, and let bi coordinates be [ bx1, by1, bx2, by2], ci coordinates be [ cx1, cy1, cx2, cy2], and di coordinates be [ dx1, dy1, dx2, dy2], the central point calculation formula is as follows:
center i x ＝{[(bx1+bx2)/2+(cx1+cx2)/2]/2+(dx1+dx2)/2}/2；
center i y ＝{[(by1+by2)/2+(cy1+cy2)/2]/2+(dy1+dy2)/2}/2；
wherein center i x Representing the abscissa of the center point, center i y Representing the center point ordinate;
if the set Pi does not contain the element bi in the set B and the element ci in the set C, but the set Pi contains the element di in the set D, the central point calculation formula is as follows:
center i x ＝(dx1+dx2)/2；
center i y ＝(dy1+dy2)/2；
if the set Pi does not contain the element di in the set D, but the set Pi contains the element bi in the set B and the element ci in the set C, the central point calculation formula is as follows:
center i x ＝[(bx1+bx2)/2+(cx1+cx2)/2]/2
center i y ＝[(by1+by2)/2+(cy1+cy2)/2]/2
if the set Pi is not the case, then the center point is empty.
In an embodiment of the present application, the method for calculating the bottom point in step S4 includes:
if the set Pi includes the element ci in the set C and the element di in the set D, and the ci coordinate is [ cx1, cy1, cx2, cy2], and the di coordinate is [ dx1, dy1, dx2, dy2], the bottom point of the set Pi is calculated according to the following formula:
bottom i x ＝[(cx1+cx2)/2+(dx1+dx2)/2]/2；
bottom i y ＝(cy2+dy2)/2；
wherein, bottom i x Representing the abscissa of the base point, bottom i y Representing the base point abscissa;
if the set Pi only contains one of the elements ci in the set C or the elements di in the set D, and the coordinates thereof are set to [ x1, y1, x2, y2], the bottom point of the set Pi is calculated according to the following formula:
bottom i x ＝(x1+x2)/2；
bottom i y ＝y2；
if set Pi is not the case, then the bottom point is empty.
In an embodiment of the present application, the method for calculating the inclination in step S4 includes:
setting the vertex coordinates as [ vx, vy ], the central point coordinates as [ cx, cy ] and the bottom point coordinates as [ bx, by ];
if the set Pi has a vertex and a center point, the inclination in is calculated by the following formula:
in＝arctan(|cy-vy|/|cx-vx|)；
wherein, | | is an absolute value, and arctan is an inverse tangent;
if the set Pi has a bottom point and a center point, the inclination in is calculated by the formula:
in＝arctan(|cy-by|/|cx-bx|)；
if the set Pi has a top point, a bottom point and a central point, the inclination in is calculated by the following formula:
in＝[arctan(|cy-vy|/|cx-vx|)+arctan(|cy-by|/|cx-bx|)]/2；
if the set Pi is not the case, the inclination in is null.
Step S5: and performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R.
In brief, the list T is first obtained by sorting. Specifically, the elements { Pi } in the set { P } are arranged in ascending order or descending order according to the size order of the horizontal axis of the Pi central point to form a list T. And then, performing intersection judgment, screening and sequencing on all elements in the list T to obtain a list R.
In an embodiment of the present application, the performing intersection judgment, screening, and sorting on the elements in the list T to obtain the list R includes:
A. creating a set Qi representing a set of all elements intersected by the element ti in the list T, and adding the ti into the set Qi;
B. sequentially judging whether the sets Q1 to Qi-1 contain ti, if a certain set Qk contains ti, deleting the set Qi, and designating Qk as the set to which ti belongs;
C. sequentially calculating the intersection ratio of the elements tj and ti in the list T, and if the intersection ratio is greater than a preset value S iou If so, the tj is considered to be intersected with the ti, and the tj is added into the set Qi or the set Qk;
D. sequentially judging elements ti in the list T, if the number of the corresponding sets Qi is smaller than a preset threshold number Sq, determining that the ti is not in one queue, and deleting the ti and the sets Qi from the list T;
E. and sorting the elements qj in the set Qi intersected with ti for any element ti in the list T, and sorting the elements qj in the set Qi in an ascending or descending manner according to the central point distance between qj and ai to obtain a sorted list R.
Specifically, an example of the operation performed on the elements ti in the list { T } in turn is as follows:
the intersection judging step comprises the following steps: first, a set { Qi } is created that represents the set where ti intersects all elements, adding ai to { Qi }.
Next, it is determined whether the sets { Q1} to { Qi-1} already contain ti. If some set { Qk } contains ti, then the set { Qi } is deleted, designating Qk as the set to which ti belongs.
Then, the intersection ratio iou of the elements tj and ti in the list T is calculated in sequence, wherein j > i. The counting formula is as follows:
iou＝A∩B/(A∪B-A∩B)；
if iou is greater than the preset threshold S iou Such as S iou At 0.1, tj is considered to intersect ti, adding tj to either { Qi } or { Qk }.
The screening method comprises the following steps: and sequentially judging elements ti in the list { T }, if the number of the corresponding { Qi } elements is less than a preset threshold number Sq, if the Sq is 3, determining that ti is not in one queue, and deleting ti and { Qi } from the list T.
The step of ordering includes: and sorting any element ti in the list { A } and the element qj in the set { Qi } intersected with ti, and sorting the elements qj in the set { Qi } in an ascending or descending order according to the central point distance between qj and ti to obtain a sorted list { Ri }.
Step S6: and (4) performing approximate straight line judgment on each element in the list R to determine the queuing list and the corresponding queuing number.
In an embodiment of the present application, the step S6 includes:
A. and carrying out neighborhood sorting on each element in the list R to judge any one or more of inclination, a central point straight line, a vertex straight line and a bottom point straight line.
In one or more embodiments, the determining the inclination in step S6 includes:
and circularly judging the element rj in the list R, if the inclination difference between rj and the set Pi is greater than a preset threshold So, if So is 30, judging that rj is not in the queue where the set Pi is positioned, and deleting rj from the list R.
In one or more embodiments, the determining the center point straight line in the step S6 includes:
connecting the center point of the element rj in the list R with the center point of rj +1 to obtain a straight line L j Connecting the center point of rj +1 with the center point of rj +2 to obtain a straight line L j+1 Calculating L j And L j+1 The included angle a is expressed by the formula:
a＝arccos[(L j *L j+1 )/(||L j ||*||L j+1 ||)]；
wherein L is j *L j+1 Dot product, | | L j And L j+1 Calculating a modulus of the | table;
if the angle a is smaller than the preset upper angle threshold Sa, if the Sa is 30, the rj +1 and the rj +2 can be connected into a straight line, and the rj, the rj +1 and the rj +2 are judged to be in a queue.
In one or more embodiments, the determining the vertex straight line in the step S6 includes:
connecting the center point of the element rj in the list R with the vertex of rj +1 to obtain a straight line L j Connecting the vertex of rj +1 with the vertex of rj +2 to obtain a straight line L j+1 Calculating L j And L j+1 The included angle b is expressed by the formula:
b＝arccos[(L j *L j+1 )/(||L j ||*||L j+1 ||)]；
wherein L is j *L j+1 Dot product, | | L j | and | L j+1 Calculating a modulus of the | table;
if the angle b is smaller than the preset angle threshold Sb, if Sb is 30, the rj +1 and the rj +2 can be connected into a straight line, and then the rj, the rj +1 and the rj +2 are judged to be in a queue.
In one or more embodiments, the determining the midpoint straight line in step S6 includes:
connecting the center point of the element rj in the list R with the bottom point of rj +1 to obtain a straight line L j Connecting the bottom point of rj +1 with the bottom point of rj +2 to obtain a straight line L j+1 Calculating L j And L j+1 The included angle c is expressed as:
c＝arccos[(L j *L j+1 )/(||L j ||*||L j+1 ||)]；
wherein L is j *L j+1 Dot product, | | L j | and | L j+1 Calculating a modulus of the | table;
if the angle c is smaller than the preset angle threshold Sc, if Sc is 30, the rj, rj +1 and rj +2 can be connected into a straight line, and then the rj, rj +1 and rj +2 are judged to be in a queue.
B. If the Sn elements in the list R meet the judgment condition of at least one of the inclination, the straight line of the central point, the straight line of the top point and the straight line of the bottom point, the list R is judged to be a queuing queue, and the number of the elements in the list R is the number of people in the queue.
Specifically, if at least one element of the list { Ri } satisfies one of the central point straight line, the top point straight line and the bottom point straight line, the list { Ri } is determined to be a queue. The resulting list { T } and sub-list { Ri }, where { Ri } may not exist. The finally existing list { Ri } is a queuing list, and the number of elements of the list { Ri } is the number of people in the queue.
In summary, the invention can realize a relatively accurate queuing counting function by using only one 2D image, can detect multiple parts of the head, the upper body, the lower body, the whole body and the like which are only partially visible when people are blocked by each other in a dense scene, can improve the detection and accuracy rate under the blocking, and further saves the manpower and cost for queuing statistics.
Referring to fig. 4, a block diagram of a multi-bit detection-based queuing statistics apparatus according to an embodiment of the present invention is shown, wherein the multi-bit detection-based queuing statistics apparatus 400 includes:
an obtaining module 401, configured to obtain and pre-process picture data;
a processing module 402, configured to detect the head, the upper body, the lower body, and the whole body of each person in the preprocessed picture data by using a detection model, represent each part by using a preset data structure including coordinate information, and acquire sets a, B, C, and D corresponding to the head, the upper body, the lower body, and the whole body, respectively; respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P; respectively calculating a vertex, a central point, a bottom point and an inclination degree corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P; performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R; and (4) performing approximate straight line judgment on each element in the list R to determine the queuing list and the corresponding queuing number.
It is understood that the apparatus 400 can implement the queuing statistics method based on multi-location detection as described in fig. 1 through the operation of each module.
It should be noted that the division of each module of the above apparatus is only a logical division, and all or part of the actual implementation may be integrated into one physical entity or may be physically separated. And these modules can all be implemented in the form of software invoked by a processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. For example, the processing module 402 may be a separate processing element, or may be integrated into a chip of the apparatus, or may be stored in a memory of the apparatus in the form of program code, and a processing element of the apparatus calls and executes the functions of the processing module 402. The other modules are implemented similarly. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element described herein may be an integrated circuit having signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.
For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as: one or more Application Specific Integrated Circuits (ASICs), or one or more microprocessors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs), among others. For another example, when one of the above modules is implemented in the form of a Processing element scheduler code, the Processing element may be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).
As shown in fig. 5, which illustrates a schematic structural diagram of a computer device in an embodiment of the present invention, as shown in the drawing, the computer device 500 includes: a memory 501, and a processor 502. The memory 501 stores a computer program, and the processor 502 executes the computer program to implement the queuing statistics method based on multi-bit detection as described in fig. 1.
The Memory 501 may include a Random Access Memory (RAM), and may further include a non-volatile Memory (non-volatile Memory), such as at least one disk Memory.
The Processor 502 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component.
In an embodiment of the present invention, a computer-readable storage medium stores a data transmission program, and the data transmission program, when executed by a processor, implements the queuing statistics method based on multi-bit detection as shown in fig. 1.
The computer readable storage medium, as can be appreciated by one of ordinary skill in the art: all or part of the steps for implementing the above method embodiments may be performed by hardware associated with a computer program. The aforementioned image processing program may be stored in a computer-readable storage medium. When executed, the program performs steps comprising the method embodiments described above; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.
These computer program programs may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
In summary, the present invention provides a queuing statistics method, apparatus, device and medium based on multi-location detection, which can effectively overcome various disadvantages in the prior art and have high industrial utilization value.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (13)
1. A queuing statistical method based on multi-position detection is characterized by comprising the following steps:
step S1: acquiring and preprocessing picture data;
step S2: detecting the head, the upper half body, the lower half body and the whole body of each person in the preprocessed picture data by using a detection model, representing each part by adopting a preset data structure containing coordinate information, and acquiring sets A, B, C and D respectively corresponding to the head, the upper half body, the lower half body and the whole body;
and step S3: respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P;
and step S4: respectively calculating a vertex, a central point, a bottom point and an inclination degree corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P;
step S5: performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R;
step S6: and (4) performing approximate straight line judgment on each element in the list R to determine the queuing list and the corresponding queuing number.
2. The queuing statistic method based on multi-part detection as claimed in claim 1, wherein the acquiring and preprocessing picture data comprises:
resetting the picture size of the acquired picture data to [3, M, N ]; wherein 3 is the number of channels after reset, M is the height after reset, and N is the width after reset; the picture data are 3 matrixes with the size of M × N, and matrix values are pixel values of the picture data;
and subtracting the respective mean value of the 3 matrixes from the pixel value, and dividing the mean value by the respective variance of the 3 matrixes to obtain the normalized picture data.
3. The multi-site detection-based queuing statistical method according to claim 1, wherein the obtaining of the detection model comprises:
marking the head, the upper half body, the lower half body and the whole body of all people in the picture respectively for each piece of picture data after preprocessing;
dividing the marked picture data into a training set and a test set according to a certain proportion;
respectively selecting a plurality of target detection algorithms, and carrying out iterative training on the picture data and the marked target part in the training set until the loss function of the training set is not reduced and the average precision mean value of the test set is not improved at the same time, so as to obtain a training model corresponding to each target detection algorithm;
and testing each training model on the test set, and selecting the training with the average precision value as a detection model.
4. A multi-site detection-based queuing statistics method according to claim 1, wherein said preset data structure is:
{class,[x1,y1,x2,y2]}；
wherein class represents a target category, takes values of 1-4 and respectively corresponds to the head, the upper body, the lower body and the whole body; [ x1, y1, x2, y2] represents the target coordinates, x1 is the abscissa of the upper left corner, y1 is the ordinate of the upper left corner, x2 is the abscissa of the lower right corner, and y2 is the abscissa of the lower right corner.
5. The multi-site detection based queuing statistical method according to claim 1, wherein the method of step S3 comprises:
step S301: respectively creating sets Pi, pk, pm and Pn to represent individuals corresponding to elements in the sets A, B, C and D, and putting the individuals into a total set P;
step S302: according to the sequence from the set D to the set A, sequentially searching for intersection part elements which satisfy a preset intersection area ratio with each corresponding individual element in the set in all the previous item sets;
step S303: moving the intersecting position elements in the previous item set to the set corresponding to the corresponding individual, and deleting the intersecting position elements from the previous item set;
step S304: and circulating the steps to obtain sets Pi, pk, pm and Pn respectively corresponding to the individuals.
6. Method according to claim 5, wherein the method of obtaining the set Pi in step S3 comprises:
step S311: creating a set Pi to represent individuals corresponding to all elements di in the set D, adding all the elements di into the set Pi, and putting the set Pi into a total set P;
step S312: sequentially calculating the intersection area of each element cj and di in the set C and dividing the intersection area by the area of cj to obtain an area ratio sj;
step S313: if the area ratio sj is larger than a preset value Sc, the cj and the di belong to the same person, the cj is added into the set Pi, and the cj is deleted from the set C;
step S314: according to the steps S312 and S313, the intersection condition of each element bj and di in the set B is sequentially calculated, bj meeting the preset intersection area ratio is added into the set Pi, and bj is deleted from the set B;
step S315: according to the steps S312 and S313, the intersection condition of each element aj and di in the set A is sequentially calculated, aj meeting the preset intersection area ratio is added into the set Pi, and aj is deleted from the set A.
7. The multi-site detection based queuing statistics method according to claim 6, wherein the method of step S313 further comprises:
if the set Pi contains a plurality of elements cj in the set C after the round-robin of the set C, selecting cj with the largest area ratio of intersection with di, deleting the rest cj from the set Pi and adding the cj back to the set C;
and/or the presence of a gas in the gas,
if a plurality of elements cj with the same intersection area ratio exist in the set Pi after the round-robin of the set C is finished, the central point distance between cj and di is calculated, the cj corresponding to the minimum central point distance is reserved, and the rest cj are deleted from the set Pi and added back to the set C.
8. The multi-site detection-based queuing statistical method according to claim 7, wherein the central point distance between cj and di is calculated as follows:
d center ＝sqrt{[(x1 cj +x2 cj )/2-(x1 di +x2 di )/2]*2+[(y1 cj +y2 cj )/2-(y1 di +y2 di )/2]*2}；
wherein x1, x2, y1 and y2 respectively represent coordinates corresponding to the elements cj and di; sqrt represents the square-on computation.
9. The queuing statistics method based on multi-part detection as claimed in claim 1, wherein the intersection judgment, screening and sorting of the elements in the list T to obtain the list R comprises:
creating a set Qi representing a set of all elements intersected by the element ti in the list T, and adding the ti into the set Qi;
sequentially judging whether the sets Q1 to Qi-1 contain ti, if a certain set Qk contains ti, deleting the set Qi, and designating Qk as the set to which ti belongs;
sequentially calculating the intersection ratio of the elements tj and ti in the list T, and if the intersection ratio is greater than a preset value S iou If so, the tj is considered to be intersected with the ti, and the tj is added into the set Qi or the set Qk;
sequentially judging elements ti in the list T, if the number of the corresponding sets Qi is smaller than a preset threshold number Sq, determining that the ti is not in one queue, and deleting the ti and the sets Qi from the list T;
and sequencing the elements qj in the set Qi intersected with ti for any element ti in the list T, and performing ascending or descending sequencing on the distances between the central points of qj and ai to obtain a sequenced list R.
10. A multi-site detection based queuing statistics method according to claim 1, wherein said step S6 comprises:
carrying out neighborhood sorting on each element in the list R to judge any one or more of inclination, a central point straight line, a vertex straight line and a bottom point straight line;
if the Sn elements in the list R meet the judgment condition of at least one of the inclination, the straight line of the central point, the straight line of the top point and the straight line of the bottom point, the list R is judged to be a queuing queue, and the number of the elements in the list R is the number of people in the queue.
11. A queuing statistics apparatus based on multi-site detection, the apparatus comprising:
the acquisition module is used for acquiring and preprocessing the picture data;
a processing module, configured to detect the head, the upper body, the lower body, and the whole body of each person in the preprocessed picture data by using a detection model, represent each part by using a preset data structure including coordinate information, and acquire sets a, B, C, and D corresponding to the head, the upper body, the lower body, and the whole body, respectively; respectively recombining and merging the sets A, B, C and D according to the intersection condition among all the parts to obtain sets Pi, pk, pm and Pn respectively corresponding to individuals, and putting all the sets into a total set P; respectively calculating a vertex, a central point, a bottom point and an inclination corresponding to each set Pi according to the coordinates of elements in each set Pi in the total set P; performing ascending or descending arrangement according to the size sequence of the horizontal axis of the central point corresponding to each set Pi in the total set P to obtain a list T, and performing intersection judgment, screening and sequencing on each element in the list T to obtain a list R; and (4) carrying out approximate straight line judgment on each element in the list R so as to determine the queuing list and the corresponding queuing number.
12. A computer device, the device comprising: a memory and a processor; the memory stores a computer program that is executed by the processor to implement the method of any of claims 1-10.
13. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211343193.1A CN115690043A (en) | 2022-10-31 | 2022-10-31 | Queuing statistical method, device, equipment and medium based on multi-part detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211343193.1A CN115690043A (en) | 2022-10-31 | 2022-10-31 | Queuing statistical method, device, equipment and medium based on multi-part detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115690043A true CN115690043A (en) | 2023-02-03 |
Family
ID=85046243
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211343193.1A Pending CN115690043A (en) | 2022-10-31 | 2022-10-31 | Queuing statistical method, device, equipment and medium based on multi-part detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115690043A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116912779A (en) * | 2023-09-12 | 2023-10-20 | 四川弘和数智集团有限公司 | Queuing detection method and device, electronic equipment and readable storage medium |
-
2022
- 2022-10-31 CN CN202211343193.1A patent/CN115690043A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116912779A (en) * | 2023-09-12 | 2023-10-20 | 四川弘和数智集团有限公司 | Queuing detection method and device, electronic equipment and readable storage medium |
| CN116912779B (en) * | 2023-09-12 | 2023-12-22 | 四川弘和数智集团有限公司 | Queuing detection method and device, electronic equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11301954B2 (en) | Method for detecting collision between cylindrical collider and convex body in real-time virtual scenario, terminal, and storage medium | |
| US8180154B2 (en) | Method and apparatus for region-based segmentation image processing using region mapping | |
| CN107871134A (en) | A kind of method for detecting human face and device | |
| US11250250B2 (en) | Pedestrian retrieval method and apparatus | |
| CN110751620B (en) | Method for estimating volume and weight, electronic device, and computer-readable storage medium | |
| CN115690043A (en) | Queuing statistical method, device, equipment and medium based on multi-part detection | |
| CN107247926B (en) | A kind of human body detecting method and device | |
| CN113892113A (en) | Human body posture estimation method and device | |
| CN114022558B (en) | Image positioning method, image positioning device, computer equipment and storage medium | |
| CN111524112A (en) | Steel chasing identification method, system, equipment and medium | |
| CN111709391A (en) | Human face and human body matching method, device and equipment | |
| CN109447006A (en) | Image processing method, device, equipment and storage medium | |
| CN110505397B (en) | Camera selection method, device and computer storage medium | |
| CN106033613B (en) | Method for tracking target and device | |
| CN107844803B (en) | Picture comparison method and device | |
| CN111369611A (en) | Image pixel depth value optimization method and device, equipment and storage medium thereof | |
| CN112862866A (en) | Image registration method and system based on sparrow search algorithm and computing equipment | |
| WO2023239799A1 (en) | Systems and methods for efficient rendering and processing of point clouds using textures | |
| CN113449540A (en) | Animal body weight detection method, device, equipment and storage medium | |
| CN115131384B (en) | Bionic robot 3D printing method, device and medium based on edge preservation | |
| CN108052755B (en) | Vector space based on completely random forest calculates intensity prediction method and system | |
| CN114913347A (en) | Clothes color identification method, system, equipment and storage medium | |
| CN115171011A (en) | Multi-class building material video counting method and system and counting equipment | |
| CN111028264B (en) | Rotation robust three-dimensional object detection optimization method and device | |
| CN110569892A (en) | Ubiquitous positioning signal static positioning method and system |
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 |