DE102020120470B4 - Method for providing an image montage, film analysis device, and data server device for operation on the Internet - Google Patents

Abstract

The invention relates to a method for providing an image montage, with a film analysis device (10) determining a media file (22) and providing playback data which, for more than one of the output sequences, indicate a user behavior of a user of the playback device (18) when playing the provided media file (22 ) describe (S2). The film analysis device (10) checks (S3) whether the playback data for one or more of the output sequences describe a contact request by the user (32) of the media file (22). If it determines that there is a request for contact, it establishes a data communication connection to a communication device (26) (S4) and determines a thematic context for the output sequence (S5). The film analysis device (10) provides an additional output sequence (S7), preferably a construction data set, which has image content in the determined thematic context, and integrates the additional output sequence into the media file (22, S8).

Description

The invention relates to a method for providing an image montage, ie a series of images or a film. In other words, the invention relates to an image assembly method.

For example, in order to sell a motor vehicle, an advertisement, an image or a video is usually created and uploaded to a vehicle sales platform. Based on the complete advertisement, an exchange usually follows via chat or a telephone call. The basis is usually the complete advertisement. In the communication usually only two parties are integrated, usually the seller and a potential buyer. Sometimes there is a Q&A section (question and answer section, comment section) based on the full listing that is visible to everyone.

Public Q&As often lose focus, i.e. which specific topic or component it is actually about. In addition, it is only possible to match the discussion points to the specific point in the ad by using your own intellectual power. In the case of telephone or chat contact, there is usually a lack of detailed and easily accessible information on the respective component on the seller's or potential buyer's side.

the U.S. 2020/0137353 A1 describes automobile sales management and video chat systems and methods using Internet site listings corresponding to a commercial enterprise's automobiles for sale.

From the U.S. 2017/0270580 A1 a data platform is known, which can include a machine learning model, which is indicative of a buyer's inclination to buy a respective motor vehicle of a plurality of motor vehicles for sale.

the U.S. 2019/0139113 A1 describes an online system that allows a buyer to find an automobile for sale and to interact with the automobile seller while remaining anonymous as long as the buyer chooses.

the EP 1 806 896 A1 describes a method for selecting and displaying at least one piece of additional information as a function of a data stream that can be transmitted from a server to a client.

The disadvantages mentioned above are not improved or not significantly improved by the prior art.

An object on which the invention is based is to enable factual and user-specific communication when using a media file on an online platform, for example.

The invention is based on the idea of checking a media file that describes, for example, a film or a series of images, for at least one scene, in particular for several or all scenes, whether the user of the media file - i.e. a viewer - has a contact request, i.e. whether he has questions and would like to contact the creator of the media file, for example. If this is the case, the film analysis device establishes a data communication connection to a communication device of the person associated with the media file. In addition, an additional output sequence is provided in the media file, for example an additional scene with more technical details or a data set with more information, for example.

The display then does not remain in its original form, but is dynamically optimized. Communication is thus very efficient and not only the current user, but every subsequent viewer or follower can immediately benefit from the information gain in the form of an update of the sequence or its arrangement or a new sequence. The quality of the content of the media file is significantly improved and the viewer, especially future viewers, has much more information that is also really interesting, which ultimately also benefits the creator of the media file, for example a seller.

In other words, the output sequence, which shows, for example, an engine compartment of a motor vehicle to be sold, can be marked (ie "tagged") automatically and/or manually, and the viewer can, for example, ask the seller a personal question. A construction model, for example, can be provided as additional information, so the viewer receives additional interesting information. The communication, for example a telephone call, can be public or private, i.e. privacy can be taken into account. The user of the media file also receives information quickly in other situations, for example if he has a question about the cost of repairing damage to the motor vehicle.

The method according to the invention for providing an image montage is carried out by a film analysis device. An image montage is a montage of static or moving images, ie a photo montage and/or a film montage. At the Bildmon Today it is preferably a film montage, i.e. a film cut, in other words the selection, processing and structuring of recorded image material in order to give a film a different form. Ideally, audio material is also processed.

The film analysis device can also be referred to as an image analysis device and can, for example, have a data memory in which a program for image analysis is stored. The film analysis can be designed, for example, as a control chip or control unit.

According to the method, the film analysis device determines a media file, ie a file that describes a series of images and/or a video film with a predetermined output order of at least two output sequences. Each of the output sequences has an image content. An output sequence describes a group of image and/or film shots that follow one another in terms of content, that is to say a series of individual images that is related in terms of content, and is therefore a film scene, for example. Preferably, the movie analyzer can determine the media file while the media file is being provided by a player or even while it is being played.

A playback device is understood to be a device component or a device group that is designed to play back media files with image content. The player can be, for example, a video device, other image output device, a screen or a data server that can play the media file on a screen where the data server is connected via the Internet, or can transmit the output content to the screen.

The film analysis device provides playback data that describe user behavior of a user of the playback device when playing the provided media file for at least one of the output sequences.

The playback data provided for all of the output sequences preferably describe a user behavior of the user. Thus, for example, the playback data may describe how many users are watching the scene, whether the user fast-forwards or aborts an output sequence, or whether the user exhibits very attentive or bored behavior, such as looking away or walking away. The behavior of the user can be recorded, for example, via body language and facial expressions, with the playback data being able to describe camera data, for example. In other words, the playback data describe a playback behavior of the user, so that it can be derived whether and/or how the respective output sequence is/was played back or is/was perceived by the user.

Alternatively or additionally, the playback data can describe an operator action by the user of the playback device, for example a touch gesture in which the user touches a touch-sensitive screen of the playback device, for example, or a voice command, for example the command “please contact us”.

The film analysis device checks whether the playback data provided for one or more of the output sequences each describe a contact request by the user of the media file. Ideally, the film analysis checks the existence of a contact request while the user is playing the respective output sequence. The contact request specifies that the user would like to contact another person assigned to the media file, for example the creator of the media file, who may be a seller of the motor vehicle that is offered in the media file.

This checking process can include, for example, checking whether the user of the media file gives a voice command “Please contact seller”, for example, presses a button provided for this purpose or performs another operating treatment that is assigned the contact function, for example a touch gesture . Alternatively or additionally, the film analysis device can analyze behavior of the user during this testing process, which can be filmed by a camera, for example, with the film analysis device being able to analyze facial expressions, for example. The corresponding camera data can then be described by the playback data. Additionally or alternatively, the playback data may also describe whether the user fast forwards, views multiple times, actively selects, or actively skips the output sequence.

If the film analysis device determines that there is a request for contact, the film analysis device can set up a data communication connection to a communication device of the person assigned to the media file, for example a mobile phone connection to a mobile device, for example a smartphone, or an Internet connection for a video chat, for example. The film analysis device also establishes a thematic context of that output sequence for which contact is desired. For example, if the user is watching a scene in which the engine compartment was filmed, the thematic context can be Example be "engine design" or "mileage data".

The film analysis device provides an additional output sequence, ie, for example, a further film scene or a data set that can show, for example, a three-dimensional model of the motor vehicle and/or a motor vehicle part. The output sequence has image content in the determined thematic context. The additional output sequence can thus be, for example, an image of a list with the mileage data or, preferably, a construction model, for example a CAD model, of the engine compartment, or for example a film scene in which individual engine parts are shown again. To provide the output sequence, the film analysis device can, for example, request the additional output sequence from the communication device of the person assigned to the media file, receive it from this device, or upload or generate an additional output sequence in the determined thematic context in its own data memory. The film analysis device can then cut the additional output sequence into the original media file via an image assembly process, in other words integrate the additional output sequence into the media file.

The advantages mentioned above result.

Depending on the respective user behavior, the film analysis device assigns an attention value to one, several or each of the output sequences described by the playback data, which value represents a degree of attention of the user to the respective output sequence. For example, if the playback data describes camera data that describes that the user looks away or walks away during the film scene, or that his eyelids droop, the film analysis device can assign a very low attention value to this scene. This can also be the case if the playback data describes that the user - or many users on that day - fast-forward or stop the movie at that movie scene. However, a high attention value can be assigned if the user or users watch the scene again or specifically select this scene or watch the scene to the end.

The film analysis device thus divides the series of images or the video film into the output sequences. These output sequences can then be evaluated depending on how the user behaves. The individual scenes are therefore not judged objectively, but user-specifically. This can optionally be coupled with a time of day, so that the attention value can be assigned to a specific time of day after an evaluation. For example, long scenes can be more popular after work in the evening because viewers have much more time to watch the whole film. Around noon, however, only those output sequences that provide a lot of information in a short and concise manner can be of interest.

When playing the output sequence whose assigned attention value meets a predetermined interest criterion, the film analysis device provides the data communication connection to the communication device of the person assigned to the media file. The user can clarify his request immediately and does not have to wait.

The predefined interest criterion can preferably specify that the assigned attention value exceeds a predefined minimum value for the degree of attention. To select the output sequences to be changed, the film analysis device can optionally use the assigned attention value to check which of the output sequences meets the predetermined interest criterion. Such “automatic tagging” can make communication between the user and the other person very efficient. The variety of the method is also available to the user if he is not familiar with the communication options or does not even know that the exemplary seller can be contacted.

The film analysis device transmits the playback data provided to a deep learning engine. A deep learning engine ("deep learning facility") is a device, device component, or program that can apply so-called deep learning (so-called deep learning, machine learning) to a variety of data. In other words, the deep learning engine is a sophisticated facility for performing deep learning, that is, an implementation of artificial intelligence. In other words, both artificial intelligence and machine learning and deep learning can be implemented using the deep learning engine. The deep learning engine can be designed and/or configured as an in-depth, artificial neural network, for example. In other words, the deep learning engine can be set up to use a machine learning method to process a large number of empirical values and/or training data, which can also be referred to as a training data set, or a data set according to a predetermined algorithm and using the large number already stored from experience to evaluate, for example via a logic contained therein, for example a correlation. This also allows further logical links to be created in the deep learning engine.

Empirical values or training data can, for example, be statistically summarized for a large number of playback data - i.e. also for a large number of different user behavior attention values.

The film analysis device operates the deep learning engine, so that the deep learning engine statistically summarizes thematic contexts for a large number of user behavior given image contents. The film analysis facility also operates the deep learning engine to process the user behavior described by the playback data provided and thereby determine an attention increase forecast. The prognosis for increasing attention includes a probability with which an additional output sequence with a thematic context will increase the user's (or a future user's) degree of attention to the media file.

If the probability of a preference for an additional output sequence exceeds a predetermined threshold value, the film analysis device operates the deep learning engine to specify an additional output sequence with the thematic context for the media file on the basis of the attention increase prediction; preferably to specify a changed output order for the changed media file.

As a result, a variety of possible film cuts is significantly increased, and the variety can also be used for a particularly situation-specific and/or user-specific purpose. The advantages mentioned above increase synergistically.

In one development, the film analysis device can describe a data record as an additional output sequence, which describes a design model of a device or object, in particular a design model of a motor vehicle and/or a motor vehicle part. Such a digital design model can be a CAD model of the motor vehicle, for example. The film analysis device can therefore preferably send a recommendation for action to the communication device, for example, which requests the creator of the video to provide a CAD model after the film analysis device has analyzed, for example, that a film scene that goes into the technical details of the engine the viewer or viewers can watch this scene again or express a contact request. If the digital construction model of this scene is then available, this is important information for the viewer and has significant added value for the viewer and the creator of the media file. The advantages become even clearer when the film analysis device recognizes that the user has tagged the output sequence, for example by pressing a button, if the film analysis device then provides the design model for this marked output sequence.

Optionally, individual output sequences can be marked or "tagged" in such a way that the user, ie the viewer, sees when viewing the output sequence that more information can be called up or requested for this scene. In such an output sequence, for example, a symbol can then be displayed that draws the user's attention to it.

In a further embodiment of the method according to the invention, the film analysis device can provide the playback data for a number of users of the playback device, preferably with the film analysis device also taking into account how often the respective output sequence is or was viewed. This improves quality of the media file per se and not just for individual viewers, and significantly reduces a likelihood that a new user will fast forward or drop scenes.

The method can be tailored very individually if the playback data provided describe a recorded operator action by the user for selecting the output sequence from a total of the output sequences of the media file. The operator action can be, for example, a touch gesture if the movie of the media file is viewed on a touch-sensitive screen or on a device with a touchpad, or, for example, a voice command. In the case of such manual “tagging”, the film analysis device can preferably be set up to evaluate the recorded operator action and to assign it to the output sequence, for example.

In one embodiment of the method, the film analysis device can include an image montage device, ie a device, a device component or a device group, which is set up to provide, ie preferably to generate, an output sequence. In this embodiment, the film analysis device can cause the image assembly device to provide the additional output sequence.

The invention also includes the film analysis device. The film analysis device can be a Have a data processing device or a processor device that is set up to carry out an embodiment of the method according to the invention. For this purpose, the processor device can have at least one microprocessor and/or at least one microcontroller and/or at least one FPGA (Field Programmable Gate Array) and/or at least one DSP (Digital Signal Processor). Furthermore, the processor device can have a program code which is set up to carry out the embodiment of the method according to the invention when executed by the processor device. The program code can be stored in a data memory of the processor device.

A film analysis device is understood to mean a device or a device component for electronic data processing, which can receive and evaluate signals and generate control signals. In particular, the film analysis device is set up for image analysis.

The film analysis device can preferably include the deep learning engine described above. The film analysis device is set up to carry out the corresponding embodiment of the method according to the invention that has already been described above and to operate the deep learning engine accordingly. However, the film analysis device can also be set up for the specified embodiment of the method according to the invention if the deep learning engine is not a component of the film analysis device but, for example, a component of a data server that is structurally separate from the film analysis device. The advantages have already been discussed above.

In a further refinement, the film analysis device can include an image montage device which is set up to provide the additional output sequence. The advantages have already been discussed above.

The above object is achieved by a storage medium with a program code which, when executed by a processor device, in particular by a processor device of a mobile terminal, is set up to cause the film analysis device to carry out an embodiment of the method described above. The storage medium can be designed, for example, as a memory card or memory chip or other data storage medium. The advantages already mentioned result.

Mobile, portable terminal with an embodiment of the storage medium described above.

The object stated above is achieved by a data server device for operation on the Internet, for example a data server, a backend and/or a data cloud, the data server device having an embodiment of the storage medium according to the invention and/or the film analysis device according to the invention.

The invention also includes developments of the film analysis device according to the invention, the storage medium according to the invention, the mobile terminal device and the data server device according to the invention, which have features as have already been described in connection with the developments of the method according to the invention. For this reason, the corresponding developments of the film analysis device according to the invention, the storage medium according to the invention, the mobile terminal device and the data server device according to the invention are not described again here.

The invention also includes the combinations of features of the described embodiments. The invention also includes implementations that each have a combination of the features of several of the described embodiments, unless the embodiments were described as mutually exclusive.

• 1 eine schematische Darstellung zu einem Ausführungsbeispiel des erfindungsgemäßen Verfahrens und der erfindungsgemäßen Vorrichtungen;
• 2 eine weitere schematische Darstellung zu dem Ausführungsbeispiel des erfindungsgemäßen Verfahrens und der erfindungsgemäßen Vorrichtungen; und
• 3 eine weitere schematische Darstellung zu dem Ausführungsbeispiel des erfindungsgemäßen Verfahrens.

Exemplary embodiments of the invention are described below. For this shows:

• 1 a schematic representation of an embodiment of the method according to the invention and the devices according to the invention;
• 2 a further schematic representation of the exemplary embodiment of the method according to the invention and the devices according to the invention; and
• 3 a further schematic representation of the embodiment of the method according to the invention.

The exemplary embodiments explained below are preferred embodiments of the invention. In the exemplary embodiments, the described components of the embodiments each represent individual features of the invention that are to be considered independently of one another and that each also develop the invention independently of one another. Therefore, the disclosure is also intended to encompass combinations of the features of the embodiments other than those illustrated. Furthermore, the described embodiments can also be supplemented by further features of the invention that have already been described.

In the figures, the same reference symbols designate elements with the same function.

the 1 illustrates the principle of the method according to the invention and the devices according to the invention according to an exemplary embodiment. This shows the 1 a film analysis device 10, which can be designed, for example, as an image processing device or a control chip. The film analysis device 10 can optionally have a processor device 12 and/or a data memory 14 . In the example of 1 For example, the film analysis device 10 can be a component of a data server device 16, which can be configured as a data server or computer, for example.

A playback device 18, for example a video device or a computer with a screen, is exemplified in FIG 1 shown as a device physically separate from the data server device 16 and the film analysis device 10. However, it is also conceivable that the playback device 18 can be part of the data server device 16, for example, or can form a device together with the film analysis device 10, for example. In the case of the structurally separate variant as in the 1 shown, the communication with the film analysis device 10 takes place via a data communication connection 20, preferably a wireless data communication connection, such as a WLAN connection, an Internet connection or a cellular connection. However, a wired data communication connection 20, for example a cable or a data bus, is also possible.

The playback device 18 can, for example, have its own storage medium 19 on which a media file 22, for example a film with a number of video scenes as output sequences, can be stored.

In the example of 1 For example, the film analysis device 10 can comprise an image assembly device 24, for example a chip or an application program for image processing. The image assembly device 24 can also be referred to as an image processing device. If the image montage device 24 is not part of the film analysis device 10, it can be, for example, an image montage device 24 of another device.

the 1 also shows another variant that can optionally be combined with the variant already mentioned, and in which the film analysis device 10 is optionally set up to generate an image montage signal and to transmit it to a receiving device 26, preferably to a data server or a mobile terminal device, for Example to a seller 40 (in the 1 not shown), i.e. to the creator of the media file 22. The image montage signal can, for example, be a recommendation to the seller 40 to provide a CAD model, or a corresponding request to the receiving device 26 to transmit such. The mobile end device can be a smartphone, a laptop or a PC, for example. Here, too, the communication can preferably take place via a wireless data communication link 21 .

In the example of 1 a deep learning engine 28 is shown as a component of the film analysis device 10 . Alternatively, however, the deep learning engine 28 can be a component of another data server (in the 1 not shown) or in the in the 1 shown data server device 16 can be integrated and preferably communicate wirelessly with the film analysis device 10.

Training data or empirical values with which such a deep learning engine 28 can have been trained can preferably be combined as an artificial neural network and come from a database, for example. Such data on user data, playback data and/or attention values in a number of >1000, in particular >10000, can preferably be/have been used to train the deep learning engine 28, with the training data preferably being recorded over a predetermined observation period . Such a data set can be referred to as a big data data set.

Each empirical value can, for example, be a combination of user behavior for image content and a corresponding thematic context for an additional output sequence, for example the combination of a detected, targeted selection of the output sequence in which the film analysis device, for example, uses a common image analysis technique to detect a fender of a motor vehicle recognized, and for example the thematic context "body". An empirical value is therefore understood to be a value or an indication that makes a statement based, for example, on empirical measurements or investigations as to whether, for example, an additional output sequence, which shows a CAD model of the body, for example, in the context of the viewed Output sequence is interesting, so the viewer's attention is increased.

The empirical value can therefore be a numerical value or an assignment value, for example. An empirical value is also understood to mean a functional dependency or a functional assignment that makes a statement about it meets whether or to which user behavior changing the respective image sequence parameter leads to an increase in the degree of attention. In other words, an empirical value can therefore also be understood as a rule for the assignment based on numerical values.

The media file 22 from the example of 1 can be, for example, a film in which the motor vehicle is offered for sale. Optionally, a large number of output sequences can be stored in the data memory 14 of the film analysis device 10 or in a data memory of the image assembly device 24, which, for example, are a CAD model of the body or the fender, but which are not yet part of the original film of the media file 22. As a result, the image montage device 24 can insert new scenes later, as in a type of modular system. A large number of sequences are preferably stored there and these can be divided into categories, for example the categories “CAD model”, “Optional equipment and accessories”, and “Price offer”.

the 2 shows a possible technical implementation in a different representation, in which the media file 22 with its output sequences A to J is represented visually. If the media file 22 is stored, for example, in a customer account and thus on a server of the motor vehicle manufacturer, eCommerce tools 29, for example, a trust provider and/or a verification layer 30 ( for example information from a test center, for example the TÜV), a "Car CV" (a kind of curriculum vitae of the motor vehicle that is to be sold with the media file 22), information about the seller (for example data from an electronic identity card), and various identification numbers can be stored. Such information is preferably only made available after corresponding activation by the consent of the user of the customer account.

In the example of media file 22 as in the 2 shown, this can initially have the output sequences A, B, C, D, E, F, G, H, I and J, which can be cut together to form a film in the playback order shown.

the 2 shows a viewer 32 schematically, for example a potential buyer who would like to buy a motor vehicle and is looking at the film stored on the playback device 18, ie the media file 22. The playback device 18 is for reasons of clarity in the 2 Not shown. Each of the output sequences A - J has an output content, i.e. image content on a specific topic that is 2 to some of the output sequences A - J as a symbol. The numbers of the respective output sequence A - I reflect an exemplary number of so-called "views", i.e. how often the respective output sequence A - I has been viewed so far. For example, the output sequence A, which introduces the seller of the motor vehicle, for example, may have been viewed 30 times. The output sequence B, which shows the exterior of the motor vehicle, can have been viewed 20 times and the output sequence C, in which the interior of the motor vehicle is shown, also 20 times. In D, for example, damage to the motor vehicle to be sold can be filmed, and this output sequence D can, for example, have been viewed 18 times. For example, E can be a scene about service maintenance, F can be a contribution to dealer offers, G can be a contribution to the driving behavior of the motor vehicle, while H can, for example, indicate contact information for feedback. For example, Output Sequence I can be a scene asking for comments and reviews. The "view" numbers of the output sequences E - I are given analogously as example numbers.

Ideally, this order of output sequences A - J can be analyzed by the deep learning engine 28 . The artificial intelligence embodied by the deep learning engine 28 can ideally be designed for learning in that it constantly updates itself with new input data. The film analysis device 10 in combination with the deep learning engine 28 can be provided in the example of the figure, for example as an influencer service. the 2 shows schematically, in order to provide an overview, a total of “followers” 34, i.e. people, institutions or profiles who use the service and have subscribed to the service, for example. This can be, for example, the manufacturer that manufactured the motor vehicle for sale, third-party data sources, dealers, a seller guide, detailing services, third-party service providers, cleaners (i.e. service providers that offer car cleaning), a workshop and a third-party artificial Intelligence. In other words, typical “followers” can be, for example, classic buy-side followers and sell-side followers.

Optionally, further data can also be included in the analysis, for example environmental data such as weather data, position data that describe where the potential viewer 32 is located, or a current time. Other data that can optionally be included are, for example, existing damage, a presentation of the person, i.e. the seller, data on different perspectives, Data for a voice analysis if the media file 22 has an audio track on which the salesperson speaks something, data for a language, for example for a language analysis, and the "views".

the 2 also shows the communication channels that can be used, for example, with optional tagging, i.e. marking. In such a tag-based communication with the seller or a follower, in particular with the seller, the potential viewer 32 can, for example, tag in the output sequence B (in the 2 illustrated by the filled circle in the output sequence B) and thus signal to the seller that he would like more details and a chat connection to the exterior.

Communication between the followers 34 and the film analysis device 10 and/or the deep learning engine 28 can take place via the sequence view, for example, via an upload and general communication.

The exemplary seller 40 can have a dashboard displayed on the receiving device 26, for example a mobile device such as a smartphone, on which he can see, for example, a recommendation for action or information from the film analysis device 10, which can be described by the optional image montage signal , preferably in combination with the possibility of viewing the media file 22, ie the “Car Story” of his motor vehicle. The individual output sequences can preferably be shown in this case. The exemplary dashboard may also provide an opportunity for video communication with the buyer 32 (or the followers), a text display for the request or recommended action, and an opportunity for operator actions, for example when operating an app using, for example, voice commands. The two arrows between the receiving device 26 and the media file 22 represent the communication between the film analysis device 10 and the receiving device 26, i.e. the transmission of the image montage signal to the receiving device 26, and the transmission of, for example, a CAD model for the output sequence B, which is uploaded by the seller 40 can be.

Optionally, the film analysis device 10 and/or the deep learning engine 28 can access a database 36, which is a database for motor vehicle identification numbers. On the basis of this vehicle identification number, a contribution of this data can then optionally flow in if there is an indication of which motor vehicle identification number the motor vehicle offered by the film has.

Analogously to this, the film analysis device 10 and/or the deep learning engine 28 can access an analytical database 38 on which, for example, the empirical values for the deep learning engine 28 can be stored. Analyzed vehicle data can therefore be stored here as a learning data set or training data set and called up when required. Such a learning data set can, for example, contain the link that, for example, experience has shown that the vehicle, model XY from year Z, has driven 10,000 kilometers. Such information can then optionally be taken into account by an additional output sequence and/or imported during a "tagging".

In the output sequence B, the uploaded CAD model can then be called up after the process has ended.

the 2 illustrates the principle of optimizing the media file 22. The representation of the 1 Fig. 12 is another representation of this, in which the devices for highlighting the process sequences are described, as explained below. That in the 1 The example shown can be used with the 2 explained features can be combined, but also stand alone form an embodiment. Instead of the receiving device 26, which is designed as a mobile terminal, the receiving device 26 of the example 1 can also be a data server, for example a workshop data server.

In a first method step S1, the film analysis device 10 determines the media file 22 that can be stored on the playback device 18 ( 1 ). The playback data are provided (S2) by, for example, receiving a corresponding signal from the playback device 18, which counts the views, for example, or detects, for example, which output sequences A-I the user presses fast forward or interrupts the scene for. The playback data can alternatively or additionally describe whether the user repeats or rewinds the output sequence.

In method step S3, the film analysis device 10 checks whether the playback data indicate that the user 32 would like to contact the seller 40. For example, camera data can be evaluated that show a particularly interested facial expression of the user 32, a corresponding operator action can be recognized, for example the voice command "contact us", or it can have been read that the user 32 watched the output sequence B, rewound it, and looked again. In the example of 1 the user 32 can post the output sequence B twice have watched each other, so that the film analysis device 10 establishes a data communication connection 21 between, for example, the playback device 18 (or another device to which the user 32 has access) and the receiving device 26 (S4).

The film analysis device 10 can determine, for example, the thematic context “vehicle exterior” as the thematic context, for example through a corresponding video signal that is linked to the output sequence B, or for example through image analysis (S5). When determining the thematic context (S5), the film analysis device can preferably operate the deep learning engine 28 for this purpose (S6), and for this purpose transmit the playback data provided to the deep learning engine 28 (S7).

The user 32 and the seller 40 can discuss the concerns of the user 32 via the data communication connection 21, which can optionally be a telephone connection. The provision of the additional output sequence, for example the CAD model, according to method step S7 can be done, for example, in that the seller 40 transmits the CAD model to the film analysis device 10, the film analysis device 10 transmits the CAD model from the film analysis device 10 from the Receiver 26 requests. In another variant, the optional image assembly device 24 of the film analysis device 10 can generate or select the CAD model. In method step S8, for example, the CAD model is then integrated into the output sequence B and marked with a magnifying glass symbol, for example, so that the user 32 knows that he can call up the CAD model by tapping on the magnifying glass symbol, for example. The CAD model, ie an additional output sequence, can be integrated by the film analysis device 10 re-cutting the film of the media file 22 .

In another example, based on the playback data provided, the film analysis device 10 can determine that the current user 32 or many viewers are watching the output sequence B twice or more, or otherwise imply that there is very likely a desire to contact them. Due to the large number of breaks in the output sequences E-I, the film analysis device 10 can allocate a very low attention value to each of these output sequences (S9). These values can be on a scale from 1 to 10, for example, with a value of 1 meaning very little attention and 10 meaning the highest possible level of attention. However, the film analysis device 10 can assign a high attention value to the output sequences B (S2). Ideally, however, the playback data for the output sequence B can describe the "tagging" or marking.

This step can preferably be performed by the deep learning engine 28 . In order to operate the deep learning engine 28 (S6), the film analysis device 10 can transmit the playback data provided to the deep learning engine 28 (S7). The processing of the provided playback data (S5) can preferably result in the provision of an attention-raising prognosis (S10).

The specified criterion of interest can be stored, for example, in the data memory 14 of the film analysis device 10 and specify, for example, that a predetermined numerical value is defined as an attention value as a threshold value and/or that the output sequence has been “tagged”. The predetermined numerical value can be a minimum value. By comparing the assigned attention values with this threshold value, it can then be checked which of the output sequences meet the specified interest criterion.

the 3 illustrates the principle of the process in a different way. In the example of 3 For example, the output sequence B may show the exterior of a motor vehicle. In a communication 42, such as a phone call, chat, or video chat, the user 32 and seller 40 can discuss additional details. According to the variants described above, an additional output sequence 44 can then be provided, which can show a CAD model, for example. In addition, for example, if an area on the engine compartment is "tagged" in the additional output sequence 44, the user 32 can call up additional information 46 in the above-mentioned manner, which, for example, describes information about a component at the front of the engine compartment can, for example, motor data. The component in the example 3 can be a motor for example.

Overall, the examples show how contribution control can be provided within a car story.

Discussions can be started by tagging based on and with reference to a specific point in time in an output sequence or even to a specific point in space (e.g. a stored 3D model, and/or a CAD model in space).

Based on the Q&A or discussion on the component and/or information within a The new knowledge is automatically added to the output sequence (ideally with the consent of the ad creator).

Among other things, it is advantageous that such a communication is as efficient as possible, and each subsequent follower can immediately benefit from the information gained in the form of an update of the sequence and/or its arrangement or a new sequence.

Claims (10)

Method for providing an image montage, wherein a film analysis device (10): - Determines a media file (22) which describes a series of images or a video film with a predetermined output sequence of at least two output sequences, each of the output sequences having image content (S1); - Provides playback data that describe (S2) a user behavior of a user of the playback device (18) when playing the provided media file (22) for more than one of the output sequences; - checks (S3) whether the playback data for one or more of the output sequences describe a contact request by the user (32) of the media file (22), the contact request specifying that the user (32) be another person assigned to the media file (22). (40) want to contact, - when playing an output sequence whose assigned attention value meets a predetermined interest criterion, and if it determines that there is a contact request: a data communication connection to a communication device (26) of the media file (22) assigned person (40) produces (S4); and determines a thematic context of that output sequence (S5) for which the contact request exists, - - on the basis of the playback data provided: one, several or each of the output sequences described by the playback data, depending on the respective user behavior, assigns an attention value (S9), which represents a degree of attention of the user to the respective output sequence; - transmits (S7) the playback data provided to a deep learning engine (28); - the deep learning engine (28) operates (S6), so that the deep learning engine (28) for a variety of user behavior given image content thematic contexts are statistically combined for an additional output sequence; - the deep learning engine (28) operates (S6) in order to process the user behavior described by the playback data provided using the deep learning engine (28) and thereby determine an attention increase forecast, wherein the attention increase forecast includes a probability with which an additional output sequence with a thematic context to an image content increases the degree of attention of the user (32) to the media file (22); and - if the probability of a preference for an additional output sequence with a thematic context exceeds a predetermined threshold value: the deep learning engine (28) operates (S6) in order to specify an additional output sequence with the thematic context for the media file based on the attention increase forecast (S5 ); preferably to specify a changed output order for the changed media file (22); - providing (S7) the additional output sequence having image content in the determined thematic context; and - The additional output sequence in the media file (22) integrated (S8). procedure after claim 1 , wherein the film analysis device (10) provides a data record as an additional output sequence that describes a design model of a device or object, in particular a design model of a motor vehicle and/or a motor vehicle part. Method according to one of the preceding claims, wherein the playback data provided describe a detected operator action by the user for selecting the output sequence from a total of the output sequences. Method according to one of the preceding claims, in which the film analysis device (10) comprises an image assembly device (24) which is arranged to provide the additional output sequence. Film analysis device (10) which is set up to carry out a method according to one of the preceding claims. Film analysis device (10) after claim 5 comprising a deep learning engine (28), wherein the film analysis device (10) is set up to: transmit the playback data provided to a deep learning engine (28) (S7); - To operate (S6) the deep learning engine (28), so that the deep learning engine (28) statistically summarizes thematic contexts for a multiplicity of user behavior given image contents for an additional output sequence; - To operate (S6) the deep learning engine (28) in order to play the data provided by the to process the described user behavior using the deep learning engine (28) and thereby to determine an attention increase forecast, wherein the attention increase forecast includes a probability with which an additional output sequence with a thematic context relates to an image content to the degree of attention of the user (32). the media file (22) increases; and - if the probability of a preference for an additional output sequence with a thematic context exceeds a predetermined threshold value: to operate (S6) the deep learning engine (28) in order to specify an additional output sequence with the thematic context for the media file based on the prognosis for increasing attention (S5); preferably to specify a changed output order for the changed media file (22). Film analysis device (10) according to one of Claims 5 or 6 comprising an image montage device (24) which is set up to integrate the additional output sequence into the media file (22) (S8). Storage medium (14) with a program code that is set up to, when executed by a processor device of a mobile terminal, a method according to one of Claims 1 until 4 to perform. Mobile, portable terminal with a storage medium (14). claim 8 . Data server device (16) for operation on the Internet, which has a film analysis device (10) according to one of Claims 5 until 7 having, and/or a storage medium (14). claim 8 .

Images