Detailed Description
The following describes in detail a specific embodiment of the information push result detection method according to the present invention with reference to the accompanying drawings.
As shown in fig. 1, fig. 1 is a flowchart of an information push result detection method according to an embodiment, and includes the following steps:
step S10, determining the time information of each playing of the detected display information in the detection time period.
In this step, the number of times a certain segment of the tested display information is played and the time information of each playing are determined within a selected period of the testing time required to be tested.
for example, in the process of circularly pushing a plurality of advertisement display information segments, within the time of selecting and detecting, a certain segment of advertisement display information is repeatedly played for a plurality of times, and the start and end time of each playing of the segment of advertisement display information is obtained, that is, the played times and the time information of each playing can be obtained.
And step S20, acquiring video clips corresponding to each playing process of the tested display information from the acquired video data according to the time information.
in the step, the collected video data in front of the playing screen is utilized, and the video clip is intercepted from the video data according to the time information corresponding to the tested display information, so that the video clip in front of the playing screen is obtained when the tested display information is played each time.
In one embodiment, the method specifically comprises the following steps:
firstly, according to the starting and stopping time in the required detection time period, intercepting the detection video data from the collected video data.
And then, video interception is carried out on the detected video message data according to the playing start and stop time of the detected display information, and video clips corresponding to each playing process of the detected display information are obtained.
For example, for building screen advertisement display information, a camera is installed on a screen, video data before the screen is played is collected, a section of detection video data for detection is intercepted from all the collected video data according to a detection time period, then a video clip is intercepted from the video data according to time information of playing the detected advertisement display information every time and with the same time value, and a video clip synchronous with the playing of the detected display information every time is obtained.
and step S30, respectively identifying the face information for watching the tested display information from each video clip.
In this step, the face information that watches the tested display information is respectively identified from the intercepted video segments by using the face identification correlation technique, so as to determine which different faces watch the tested display information in each playing process.
In one embodiment, the identification process of step S30 may include the steps of:
Step S301, detecting a face frame picture containing a face from the video clip by a face detection technology.
Specifically, the video segments are decoded to obtain frame pictures, a face detection algorithm is used for carrying out face detection on images of each decoded frame picture, a face tracking algorithm is used for tracking the detected face, the face frame pictures containing the face are detected, and the face frame picture where each face is located and time information of the face frame picture are recorded.
step S302, detecting eye information in the face frame picture through an eye detection technology; specifically, in each frame of face frame picture, eye information is found by using a five sense organs positioning algorithm.
And step S303, acquiring a face image for watching the detected display information in each frame of face frame picture according to the eye information.
specifically, the eye information in the face frame picture is filtered, the eyes for watching the tested display information are screened out by judging the eyeball direction, and the face images and the time information thereof corresponding to the eyes are reserved.
and step S304, acquiring the face information for watching the tested display information from the watching face in the video clip.
specifically, for each section of the detected display information, based on the obtained face image, in each playing time range of the section of the detected display information, face images with overlapped time and overlapped time information are screened from the face images, and all face information for watching the detected display information and the watching time of each face are obtained.
For example, in the push result detection of the building screen advertisement display information, the collected video is decoded, and a human face detection algorithm is used for performing human face detection on each decoded frame of image. And tracking the detected face by using a face tracking algorithm, and recording all face frame pictures and time information of the face. For each tracked face, eye information is found in each frame of face frame picture by using a five sense organs positioning algorithm, and then whether eyes watch a screen or not is judged according to the direction of eyeballs. And filtering out frame pictures of which eyes do not watch the screen to obtain all the faces watching the screen and the time information of each face watching the screen.
And step S40, clustering the face information to obtain the information of the viewer who actually views the tested display information.
in this step, the face information obtained in the process of playing the detected display information each time is obtained in the above steps, all the face information in the detection time period is subjected to clustering processing, and the face information in the detection time period which has watched the detected display information is obtained, so that the character information of the viewer who actually watches the detected display information can be determined.
in one embodiment, the clustering process of step S40 may include the following steps;
Step S401, extracting a plurality of frames of face images from face frame pictures corresponding to each piece of face information; the extraction interval can be set according to actual conditions.
And step S402, marking the same ID number on the face image of the same face in the same video clip.
specifically, in the same video segment, the face images of the same face in each frame of face picture are labeled with the same ID number, so that different face images have different ID numbers, and the face images of the same face also have different ID numbers in different video segments.
step S403, clustering the face images extracted from all the video segments, combining the faces with the same ID number in different categories to obtain the categories of all the faces, and then determining the information of the viewers who actually view the tested display information according to the categories of the faces.
Specifically, all face images are clustered, because faces of the same viewer may appear in different video segments, a plurality of personal face image categories are obtained after clustering, each category corresponds to a face of the viewer, and in the clustering process, face images of the same viewer may also appear in different categories, so that the face images recorded by the ID numbers are merged if the face images of different categories have the same ID, and the categories are considered as the face images of the same viewer. The categories of all the faces are obtained after combination, each category corresponds to one viewer, and therefore the viewer information of the actually viewed display information can be determined, preferably, the viewer information comprises the number of the viewers, the viewing times of each viewer, the time length of each viewing and the like.
For example, in the above-mentioned pushing result detection of the building screen advertisement display information, all faces are clustered by using a clustering algorithm based on the face information of a specific one of the advertisement display information. Because the advertisement display information can be watched by the same person for multiple times, through clustering processing, the number of different persons who actually watch the advertisement display information can be known, and the times of watching the advertisement display information by each person and the time of watching each time are known.
In order to achieve a better clustering effect, each face extracts a plurality of face images for clustering, one face is selected every M frames in a time range for watching the face for one time of advertisement display information playing, if the time length for watching the advertisement display information is T, the extracted face images have N = T/M total number, the N faces are labeled with the same ID number, and the face images selected by different faces are labeled with different ID numbers. Clustering all the labeled face images to obtain a plurality of categories, if the face images in different categories of faces have the same ID number, indicating that the face images belong to the same face, combining the categories, and after combination, obtaining n categories, thereby determining that n viewers actually watch the advertisement display information in the detection time period, knowing the times of watching the advertisement display information by each viewer in the detection time period according to the times of the faces of each category appearing in the video segment, and obtaining the time of watching each time according to the time information recorded in the face images.
the technology of the invention can accurately detect the push result of the display information, including face information, watching times, watching time length and the like of a viewer, and can execute the application related to the display information by using the detected push result. For example, the method is used for counting the advertisement effect, and further providing reference and modulation strategy support for advertisement display information pushing, and in addition, the method can also be applied to occasions such as conferences and training, and the like, and is used for detecting the display information pushing result, and acquiring information such as the attention of participants to the display information and the acceptance degree of trainees to the display information in training.
the following describes in detail a specific embodiment of the information push result detection system according to the present invention with reference to the drawings.
As shown in fig. 2, fig. 2 is a schematic structural diagram of an information push result detection system according to an embodiment, and the schematic structural diagram includes:
The time information acquisition module is used for determining the time information played by the tested display information in each time within the detection time period;
the video clip intercepting module is used for acquiring video clips corresponding to the playing processes of the tested display information from the acquired video data according to the time information;
the face information identification module is used for identifying the face information for watching the tested display information from each video clip;
And the viewer information acquisition module is used for clustering the face information to acquire viewer information for actually watching the tested display information.
In one embodiment, the viewer information includes: the number of viewers, the number of views per viewer and the length of time of each view.
In one embodiment, the video clip intercepting module comprises:
the detection video data intercepting unit is used for intercepting the detection video data from the collected video data according to the starting and stopping moments in the required detection time period;
And the video clip intercepting unit is used for carrying out video interception on the detection video message data according to the playing starting and ending time of the detected display information to obtain video clips corresponding to each playing process of the detected display information.
In one embodiment, the face information recognition module includes:
The face detection unit is used for detecting a face frame picture containing a face from the video clip by a face detection technology;
The eye information detection unit is used for detecting the eye information in the face frame picture through an eye detection technology;
the face image acquisition unit is used for acquiring a face image for watching the detected display information in each frame of face frame picture according to the eye information;
And the face information acquisition unit is used for acquiring the face information for watching the tested display information from the face image in the video clip.
in one embodiment, the viewer information acquiring module includes:
The image extraction unit is used for extracting a plurality of frames of face images from the face frame images corresponding to each piece of face information;
the ID number labeling unit is used for labeling the same ID number of the face images of the same face in the same video clip;
the clustering processing unit is used for clustering the face images extracted from all the video segments and combining the faces with the same ID number in different classes to obtain the classes of all the faces;
And the viewer information determining unit is used for determining the viewer information for actually watching the tested display information according to the type of the human face.
the information push result detection system and the information push result detection method of the present invention correspond to each other one to one, and technical features and advantageous effects thereof described in the embodiments of the information push result detection method are applicable to the embodiments of the information push result detection system, which is described here.
it will be understood by those skilled in the art that all or part of the processes of the above embodiments and the corresponding systems may be implemented by a computer program, which can be stored in a computer-readable storage medium, and when executed, can include the processes of the above embodiments. The storage medium may be a magnetic disk, an optical disk, a Read-only Memory (ROM), a Random Access Memory (RAM), or the like.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.