CN110955799A - Face recommendation movie and television method based on target detection - Google Patents
Face recommendation movie and television method based on target detection Download PDFInfo
- Publication number
- CN110955799A CN110955799A CN201911180954.4A CN201911180954A CN110955799A CN 110955799 A CN110955799 A CN 110955799A CN 201911180954 A CN201911180954 A CN 201911180954A CN 110955799 A CN110955799 A CN 110955799A
- Authority
- CN
- China
- Prior art keywords
- television
- face
- file
- interface
- target detection
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/73—Querying
- G06F16/735—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/23—Updating
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/51—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Abstract
The invention discloses a face movie recommendation method based on target detection, relates to the technical field of intelligent televisions, and aims to solve the problem that in the prior art, when the user is not aware of the star identity in a television interface, more related movie information cannot be conveniently inquired and obtained. The technical scheme is characterized in that deep learning of images is carried out at a server side, the image deep learning is applied to a television side, and face images of a television side interface are detected and matched, so that the star identity is obtained. The invention can quickly identify the star in the television interface through the target detection algorithm under the condition that the star information of the user on the television interface is ambiguous, present the video information related to the star for the user, provide the latest video works for the user at the first time and realize personalized recommendation.
Description
Technical Field
The invention relates to the technical field of smart televisions, in particular to a face recommendation movie and television method based on target detection.
Background
The smart television has successfully occupied the mainstream of the market, and the main functions of most televisions can be basically summarized as follows: searching movies and videos, recommending educational resources, inquiring weather information, setting alarm clock reminding and the like. The main function of the television is still video entertainment so far, so that if personalized recommendation of television videos can be realized, better experience can be brought to a television user group.
In the prior art, only one or two labels of the lead actor are marked in the video, for most of the matchers, users cannot know their names or other works through the video brief introduction, and for users who want to continuously check the works of the matchers, the users can continuously watch the works on the television by searching the video names through the television after searching by using other network tools is completed, so that the personalized experience of the users is seriously influenced.
Therefore, if the information of the actor can be acquired according to the video image of the television terminal interface in the intelligent television and other works of the actor can be automatically searched, the user experience can be greatly improved.
Disclosure of Invention
The invention aims to provide a face movie recommendation method based on target detection, which has the effects that under the condition that the star information of a user on a television interface is ambiguous, stars in the television interface are quickly identified through a target detection algorithm, and all the video information of the stars can be presented to the user.
The above object of the present invention is achieved by the following technical solutions:
a face recommendation movie and television method based on target detection comprises an intelligent television end and a server end, and is characterized by comprising the following steps:
s1, labeling the face photos by adopting a picture labeling tool at the server end to generate a labeling file, wherein the labeling file contains label information representing the identity of the face photos;
s2, converting the label file into a tfrecrd format file;
s3, training the model by using the tfrecrd format file to obtain a target detection training model, and converting the target detection training model into a tflite file;
s4, compiling the tflite file into an apk file, and installing the apk file at the television end;
s5, detecting the face image of the television interface at the television, matching the face image by using the apk file, calling the label information corresponding to the matched face photo, and displaying the face photo identity represented by the label information on the television interface.
The invention is further configured to: when the label information is called, the label information is uploaded to a server side to search for related film and television works, and all the inquired film and television works are sequentially displayed in a television side interface.
The invention is further configured to: and when the face image matching of the television terminal interface fails, the television terminal interface displays prompt information.
The invention is further configured to: and when the number of the results matched by the face images of the television terminal interface is more than one, sequentially arranging and displaying all the results.
The invention is further configured to: and automatically acquiring the face pictures in the network by using a crawler technology, processing the face pictures according to the rules of the steps S1-S3, adding the processed tflite file into the apk file, and automatically updating the face image database.
The invention is further configured to: and pausing or selecting screen capture to obtain the face image of the television.
The invention is further configured to: in step S5, detecting, at the television side, that the face image of the television-side interface is detected and matched after the face image of the television side is acquired by pausing or screen capturing, the face image of the television side interface or the screen capturing image.
The invention is further configured to: in step S5, detecting the face image of the television-side interface at the television side is to perform detection matching on the interface in real time, and when a pause or screen capture operation occurs, displaying the identity of the detected face photo at the television-side interface.
In conclusion, the beneficial technical effects of the invention are as follows:
1. firstly, converting an image into a tfrecrd file, which is beneficial to accelerating training, simplifying image label matching and realizing data set training, then using a model file trained by TensorFlow on a desktop PC or a server, only converting the model file into a tflite file by using an off-line tool, directly using the tfield file to operate on TensorFlow Lite, and finally realizing TensorFlow model deployment on a television based on TensorFlow Lite, thereby detecting stars at the television and individually recommending movie and television works;
2. the face image is acquired by utilizing the crawler technology, the image is processed, and the face image database is updated, so that recognizable stars are continuously added, and the video recommendation can meet the user requirements.
Drawings
Fig. 1 is an overall structural block diagram of an embodiment of the present invention.
Detailed Description
Referring to fig. 1, the invention discloses a face recommendation movie and television method based on target detection, which comprises an intelligent television end and a server end, and comprises the following steps:
s1, labeling the face photos by adopting a picture labeling tool at the server end to generate a labeling file, wherein the labeling file contains label information representing the identity of the face photos;
s2, converting the label file into a tfrecrd format file;
s3, training the model by using the tfrecrd format file to obtain a target detection training model, and converting the model into a tflite file;
s4, compiling the tflite file into an apk file, and installing the apk file at the television end;
s5, detecting the face image of the television interface at the television, matching the face image by using the apk file, calling the label information corresponding to the matched face photo, and displaying the face photo identity represented by the label information on the television interface. When the label information is called, the label information is uploaded to a server side to search for related film and television works, and all the inquired film and television works are sequentially displayed in a television side interface.
Wherein, the labeling tool can be LabelImg, Labelme, yolo _ mark, Vatic, Sloth, Annotrioeus, RectLable, VoTT, etc. The compiling of the apk file in steps S1 to S3 and S4 is completed at the server side.
Under normal conditions, the training folders often generate thousands of files such as pictures or texts, and the files are stored in a hashed mode, so that not only is the disk space occupied, but also the files are very slow and cumbersome to read one by one, and a large amount of memory space is occupied. the tfrecord format file storage would reasonably store data. the tfrechord internally uses a Protocol Buffer binary data coding scheme, only occupies one memory block, only needs a mode of loading a binary file at one time, is simple and quick, and is particularly friendly to large-scale training data.
When the facial image matching is performed, it may happen that the image of the television interface is not a face and cannot be recognized, or the facial image is not recorded in the database, at this time, the facial image matching of the television interface fails, the television interface displays prompt information, and accordingly, the user is advised to reselect the interface or continue playing. The star may affect the detection matching during makeup, and may have a plurality of matching results. And when the number of the results matched by the face images of the television terminal interface is more than one, sequentially arranging and displaying all the results. In this embodiment, the ranking may be performed by the heat of the stars, or may be performed by the similarity of the matching.
And automatically acquiring the face pictures in the network by using a crawler technology at a server, processing the face pictures according to the rules of the steps S1-S3, adding the processed tflite file into the apk file, and automatically updating the face image database, so that the video recommendation can meet the user requirements.
The remote controller of the television end pauses or selects screen capture, so that the television end boundary stays on a certain interface, the face image of the television end is obtained, and detection matching and movie recommendation are conveniently and accurately performed.
In step S5, detecting, at the television side, that the face image of the television-side interface is detected and matched after the face image of the television side is acquired by pausing or screen capturing, the face image of the television side interface or the screen capturing image.
In addition, the detection of the face image of the television terminal interface at the television terminal can also be realized by detecting and matching the interface in real time, when the television terminal acquires that the user pauses or performs screen capture operation, the identity of the face image detected and matched at the moment can be rapidly displayed on the television terminal interface, and related film and television works can be inquired to form a recommendation list.
The implementation principle of the above embodiment is as follows: the target detection and movie recommendation are combined, and the advantages of the target detection are utilized, so that the experience of a television user is greatly improved. The face of the television picture can be detected in real time at the television end through the trained target detection model, and the obtained face label is transmitted to the background server, so that corresponding movie and television information is retrieved, the movie and television works of the star are presented to the user finally, and personalized recommendation is realized.
The embodiments of the present invention are preferred embodiments of the present invention, and the scope of the present invention is not limited by these embodiments, so: all equivalent changes made according to the structure, shape and principle of the invention are covered by the protection scope of the invention.
Claims (8)
1. A face recommendation movie and television method based on target detection comprises an intelligent television end and a server end, and is characterized by comprising the following steps:
s1, labeling the face photos by adopting a picture labeling tool at the server end to generate a labeling file, wherein the labeling file contains label information representing the identity of the face photos;
s2, converting the label file into a tfrecrd format file;
s3, training the model by utilizing the tfrecrd format file to obtain a target detection training model, and converting the target detection training model into a tflite file;
s4, compiling the tflite file into an apk file, and installing the apk file at the television end;
s5, detecting the face image of the television interface at the television, matching the face image by using the apk file, calling the label information corresponding to the matched face photo, and displaying the face photo identity represented by the label information on the television interface.
2. The method for recommending movies based on the faces of the targets as claimed in claim 1, characterized in that: when the label information is called, the label information is uploaded to a server side to search for related film and television works, and all the inquired film and television works are sequentially displayed in a television side interface.
3. The method for recommending movies based on the faces of the targets as claimed in claim 2, characterized in that: and when the face image matching of the television terminal interface fails, the television terminal interface displays prompt information.
4. The method for recommending movies based on the faces of the target detection as claimed in claim 3, characterized in that: and when the number of the results matched by the face images of the television terminal interface is more than one, sequentially arranging and displaying all the results.
5. The method for recommending movies based on the faces of the target detection as claimed in claim 4, characterized in that: and automatically acquiring the face pictures in the network by using a crawler technology, processing the face pictures according to the rules of the steps S1-S3, adding the processed tflite file into the apk file, and automatically updating the face image database.
6. The method for recommending movies based on the faces of the targets as claimed in claim 1, characterized in that: and pausing or selecting screen capture to obtain the face image of the television.
7. The method for recommending movies based on the faces of the target detection as claimed in claim 6, characterized in that: in step S5, detecting, at the television side, that the face image of the television-side interface is detected and matched after the face image of the television side is acquired by pausing or screen capturing, the face image of the television side interface or the screen capturing image.
8. The method for recommending movies based on the faces of the target detection as claimed in claim 6, characterized in that: in step S5, detecting the face image of the television-side interface at the television side is to perform detection matching on the interface in real time, and when a pause or screen capture operation occurs, displaying the identity of the detected face photo at the television-side interface.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911180954.4A CN110955799A (en) | 2019-11-27 | 2019-11-27 | Face recommendation movie and television method based on target detection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911180954.4A CN110955799A (en) | 2019-11-27 | 2019-11-27 | Face recommendation movie and television method based on target detection |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110955799A true CN110955799A (en) | 2020-04-03 |
Family
ID=69978542
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911180954.4A Pending CN110955799A (en) | 2019-11-27 | 2019-11-27 | Face recommendation movie and television method based on target detection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110955799A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113361353A (en) * | 2021-05-28 | 2021-09-07 | 同济大学 | Zebrafish morphological scoring method based on DeepLabV3Plus |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107205172A (en) * | 2016-03-18 | 2017-09-26 | 百度在线网络技术(北京)有限公司 | A kind of method and device that search is initiated based on video content |
CN109151597A (en) * | 2018-09-03 | 2019-01-04 | 聚好看科技股份有限公司 | information display method and device |
-
2019
- 2019-11-27 CN CN201911180954.4A patent/CN110955799A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107205172A (en) * | 2016-03-18 | 2017-09-26 | 百度在线网络技术(北京)有限公司 | A kind of method and device that search is initiated based on video content |
CN109151597A (en) * | 2018-09-03 | 2019-01-04 | 聚好看科技股份有限公司 | information display method and device |
Non-Patent Citations (1)
Title |
---|
熊亚蒙: "基于TensorFlow的移动终端图像识别方法" * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113361353A (en) * | 2021-05-28 | 2021-09-07 | 同济大学 | Zebrafish morphological scoring method based on DeepLabV3Plus |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110781347B (en) | Video processing method, device and equipment and readable storage medium | |
CN111611436B (en) | Label data processing method and device and computer readable storage medium | |
CN103606310B (en) | Teaching method and system | |
US20210314635A1 (en) | System and method for providing image-based video service | |
US20090316961A1 (en) | Method for tagging image content | |
CN109688475B (en) | Video playing skipping method and system and computer readable storage medium | |
US11630862B2 (en) | Multimedia focalization | |
US10789474B2 (en) | System, method and apparatus for displaying information | |
US9906840B2 (en) | System and method for obtaining information relating to video images | |
CN103686344A (en) | Enhanced video system and method | |
CN104184923A (en) | System and method used for retrieving figure information in video | |
CN109408672B (en) | Article generation method, article generation device, server and storage medium | |
CN104837059A (en) | Video processing method, device and system | |
US10628955B2 (en) | Information processing device, information processing method, and program for identifying objects in an image | |
CN111046197A (en) | Searching method and device | |
CN109600625B (en) | Program searching method, device, equipment and medium | |
CN112612690A (en) | User interface information processing method and device, electronic equipment and storage medium | |
CN113395605A (en) | Video note generation method and device | |
CN110750719A (en) | IPTV-based information accurate pushing system and method | |
WO2019146466A1 (en) | Information processing device, moving-image retrieval method, generation method, and program | |
CN110955799A (en) | Face recommendation movie and television method based on target detection | |
KR101843355B1 (en) | Video processing apparatus using qr code | |
CN112328834A (en) | Video association method and device, electronic equipment and storage medium | |
CN116049490A (en) | Material searching method and device and electronic equipment | |
JPH10254901A (en) | Method and device for retrieving image |
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 |