System and method for collaborative image selection
A system and method for sharing images and collaborating in the selection of images likely to be interest to a user. Images (e.g., photos) are shared within a community of users, by allowing user to access any unrestricted community image, recommend an image to another user, assemble compilations of any images the user can access, etc. Various metrics regarding user activity are tracked, such as how often an image was viewed, downloaded, recommended, included in a compilation, printed, edited, etc. The metric values may be normalized, and then weighted and combined to produce, for each image, a ranking or rating personalized to an individual user. Ratings of images for a user may be used to order them for electronic presentation, assemble a set of images for an electronic compilation (e.g., an online album) or physical compilation (e.g., a photo book, a yearbook) or for some other purpose.
Latest Picateers, Inc. Patents:
- System and method for automated layout of collaboratively selected images
- Method and Apparatus for Assembling a Targeted Set of Photographs
- System and Method for Automated Layout of Collaboratively Selected Images
- System and method for collaborative image selection
- System and method for collaborative image selection
This application is a continuation of U.S. patent application Ser. No. 11/438,948, filed May 23, 2006 and issued ______ as U.S. Pat. No. ______, which claims priority to U.S. Provisional Patent Application No. 60/683,573, which was filed May 23, 2005 and is incorporated herein by reference.
BACKGROUNDThis invention relates to the field of computer systems. More particularly, a system and methods are provided for selecting and suggesting images (e.g., photographs) a person may find interesting or enjoyable, based on ratings, recommendations and other actions taken by a community of users having access to the images.
The popularity of digital imagery (e.g., digital photography, digital image processing software) has naturally led to the creation and sharing of a plethora of digital images. As with traditional film images, however, some issues and problems become more apparent, and more difficult to solve, as the number of images increases and as the images become available to more people.
For example, there is always the basic problem of finding a particular image. This problem may involve finding a specific image among hundreds, thousands or even millions of images, or selecting an image appropriate for a given situation or use (again, from hundreds or thousands of candidate images). One person searching for a specific or suitable image faces a daunting task when attempting to sift through so many images, even when they can be quickly reviewed on his or her computing or communication device (e.g., computer, telephone).
Further, sharing images between people takes more time and effort as the number of images grows. If each image is to be shared with multiple people, the problem becomes even larger. Making any image available to any person gives everybody free access, but increases the number of images they must examine in order to find interesting or enjoyable ones.
Yet further, a person wishing to assemble a photo book or other compilation of digital images may face a daunting task not only in selecting images for inclusion in the compilation, but also in arranging the images within the compilation. She naturally desires to select the most interesting images, from her standpoint or from the view of someone the book is to be given to. But, as described above, she may face a daunting task in reviewing all images available to her.
SUMMARYA system and method are provided for sharing images and collaborating in the selection of a set of images likely to be interest to a user.
In one embodiment, images (e.g., photos) are shared within a community of users, and possibly between user communities. A user may access any image not otherwise restricted, and can recommend images to other users, assemble compilations of any images the user can access, order a physical compilation (e.g., a photo book, a yearbook) etc.
As users view images, make recommendations, explicitly rate images (e.g., on clarity, composition, interest), download images, assemble compilations, and/or take other actions, metrics regarding their activities are tracked. The metric values may be normalized, and then weighted and combined to produce, for each image and for each user, a ranking or rating of the estimated level of interest that the image may hold for the user.
Ratings of images for a user may be used to order them for electronic presentation to the user, assemble a set of images for an electronic compilation (e.g., an online album) or physical compilation, or for some other purpose.
DESCRIPTION OF THE FIGURES
The following description is presented to enable any person skilled in the art to make and use the invention, and is provided in the context of particular applications of the invention and their requirements. Various modifications to the disclosed embodiments will be readily apparent to those skilled in the art and the general principles defined herein may be applied to other embodiments and applications without departing from the scope of the present invention. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.
In an embodiment of the invention, a system and method are provided for collaborating on the selection, sharing and management of images. Although embodiments of the invention may be described with reference to digital photographs, such embodiments may be readily modified for other types of images (e.g., cartoons, pictures, charts), whether still or moving.
The images may exist in different forms (e.g., digital, hardcopy) at different times, depending on the action being taken. For example, a set of images may be created and shared in digital form for easy storage and transmission via electronic mail, file sharing and so on. However, some or all of those images may be printed or otherwise embodied in a physical compilation such as a photo book, yearbook, catalog, etc. Embodiments of the invention described herein are well suited for selecting appropriate images to be included in a physical compilation.
In an embodiment of the invention, users may be grouped into communities that are defined or differentiated by any number of characteristics, and may collaborate in different ways or capacities at different times. Illustratively, a community may be defined by a common interest or activity of its users. For example, a community of users may be organized around a community organization, a geographical location, a sport or hobby, a particular school that a family member of the users (e.g., son, daughter, nephew, niece, grandchild) attends, etc.
A community may include any number of users, and a user may belong to any number of communities. A user may participate in a community by creating images (e.g., taking digital photographs), distributing them, processing (e.g., “touching up”) images, storing images, evaluating or recommending images (e.g., by alerting other users to images that may be of interest to them), arranging images for printing or other presentation, and so on. Any particular image may be shared among any number of communities, and among all or a subset of users within a particular community.
A community may be nested within another community. For example, a community associated with a kindergarten class may share images relating to students and activities of the class. This kindergarten community may be a sub-community associated with multiple grade levels, such as a lower school or grade school. Nesting of communities may be extended any number of levels. Thus, the lower school community may be part of a community associated with a school group that also includes an intermediate school, a high school and so on.
Images may be shared automatically. In other words, a community user or member need not explicitly indicate who an image should be shared with. Instead, images may be automatically shared throughout a community (including future members) and/or across communities. However, policies may be created to identify which images should or should not be shared, who they should or should not be shared with, etc. Therefore, one user may specify that some set of images should be shared with a specific user, or that all images he creates that match certain criteria (e.g., date, event, subject) should automatically be shared with some or all community members.
Images may be organized, stored and shared individually (e.g., as separate files) and/or as compilations (e.g., electronic albums, books), and may be shared among any number of users and communities. A compilation may comprise a set of images having some common theme (e.g., an event, a subject) and/or a relatively loose collection of images that have little or no inter-relation. As will be described in more detail below, a compilation may be created, organized or reorganized manually and/or automatically, and may be used to generate a hardcopy book (e.g., photo book, yearbook) with manual or automated layout (e.g., image sizing, arrangement, background, cover art).
Any image may be included in any number of compilations, and in any number of communities. And, one compilation may contain (or be contained within) any number of other compilations. References herein to compilations (e.g., for purposes of organizing or sharing images) may be interpreted to cover virtually any form or type of collection of images (e.g., slideshows, photo books, yearbooks, folders of images, any set of images transmitted or stored in logical or physical proximity).
An image may be annotated by a user or may be accompanied by metadata that includes various information regarding the image, such as date created, creator, location, caption, inscription, keywords, tags, description (e.g., of who or what is in the image), who (e.g., what users) have recommended or not recommended the image, who likes or does not like the image, etc.
In an embodiment of the invention, an image collaboration system described herein is employed to promote collaboration between users. More particularly, a user may access an image and take various actions, such as recommend it to another user, forward it (a copy of the image) to another user, add it to a compilation of images, rate or rank it on how interesting or relevant it is (e.g., to the viewing user, to another user), submit a caption, description or other metadata, and so on.
An estimate of how interesting an image may be for a user or a group (e.g., community or sub-community) of users may be based on a variety of criteria or factors mentioned herein. Some of those factors include how often an image is viewed by a given user and/or a set of users, the size with which it is viewed, how often it is included in a compilation of images assembled for a user, how often a hard-copy version of the image was requested, how frequently it was downloaded or forwarded, and so on.
The system can then automatically select images for presentation to a given user by applying what it has learned about the images. For example, if multiple users have recommended an image to the given user, when that user accesses the system to view images the recommended image may be assigned a relatively high rating in terms of how interesting or relevant it probably is to that user. An image's rating may determine whether it is presented to the user, the order in which it is presented, and/or other actions.
Further, relationships between users can be learned and applied to affect image rankings. For example, recommendations of a particular image to a given user may carry more weight if they are from relatives of the user or from other users that the given user interacts with (e.g., exchanges images and/or recommendations with) than if they are from people the user has more tenuous relationships with.
Community-level activity may also affect how interesting or relevant an image is determined to be for a user. For example, if an image is released in multiple communities, but is accessed or is more frequently recommended in a first community than a second community, it may be deemed more interesting to users in the first community and may be presented to them more often than to users in the second community.
Thus, in different embodiments of the invention, various metrics described herein may be employed to determine how interesting an image is to a specific user, a community of users and/or a subset of users with a community (e.g., users within one family). Images may be assigned values for any number of metrics, and those values may be weighted by some determination of how closely they correlate with user interest.
The rating or ranking of an image may be affected by its grouping with another image. For example, if a user adds an otherwise relatively low ranking to an image comprising relatively high ranking images, or arranges the image on a page (e.g., of an electronic album, of a layout for a physical book) with a higher ranking image, the ranking of the image may be increased.
By calculating and combining the weights in some fashion, images can then be ranked based on their estimated relevancy. Separate calculations may be done for each user, for each of any number of images. The values finally assigned to the images may be used not only to determine whether and when to present them to a user for review, but may be used for other purposes as well.
For example, in one embodiment of the invention the interest rankings of images may be used to facilitate the automatic generation of a compilation of images for an electronic album and/or a physical image book such as a photo book or yearbook. In particular, a user may be able to submit a single request, through just a single click of a mouse, for example, and the system can automatically select a default or selected number of images and offer to have the images printed and bound into a book. The selected images may, of course, be used for some other purpose, such as an electronic image album, a slideshow, wallpaper or background for a user's computer display device, etc.
When a set of images is selected for a compilation, whether electronic or physical in nature, the images may be laid out automatically by the system and/or manually by a user. For example, the system may attempt to arrange the images in chronological order, and may also attempt to maximize the compilation's appeal by considering an image's resolution or size, the orientation of an image (e.g., portrait or landscape), a theme of the compilation (e.g., a school field trip, a party), etc.
Users 102 may create images in different ways, through digital or analog photography using a stand-alone camera or a camera embedded in a telephone, with graphics software, hand drawing and so on. In this embodiment, however, each image is captured in a digital form at some point, using virtually any protocol or file structure now known or hereafter developed, such as bitmap (.bmp), windows metafile or enhanced windows metafile (.wmf, .emf), jpeg (.jpe, .jpeg, .jpg), portable network graphics (.png), etc.
The collaboration system allows users to share images with other users, recommend images to users, organize images into compilations (and share the compilations), request generation of an electronic compilation (e.g., based on user recommendations or rankings), request creation of a physical image book (e.g., a photo book, a yearbook), and take other actions with regard to the images, as described below.
In particular, the collaboration system may facilitate discussion of an image, through annotations to the image's metadata, online text or voice chat, electronic mail, etc. Discussion regarding a particular image (or compilation of images) may affect the system's ranking of that image (or compilation) for the communicating users and/or other users.
An image collaboration system such as the system of
Central server 120 comprises web server 124 for interfacing with users via their computing/communication devices (e.g., laptop computers, desktop computers, smart phones, personal digital assistants). Web server 124 thus receives requests from users and serves responses to their requests, and may perform authorization and/or authentication functions.
In the illustrated embodiment of the invention, image application server 130 is responsible for facilitating collaboration, recommendation, ranking, organization and other manipulation of images. In one implementation of this embodiment, the image application server employs the Struts web application framework. This framework is deployed within a servlet container (e.g., as defined by the Java Servlet Specification), such as a container provided by Apache Tomcat. Struts enables the separation of user interface functions from image processing or handling logic (e.g., the “business logic”).
In this implementation, the Struts framework includes an ActionServlet designed to receive a request (e.g., from a web server, directly from a user's web browser) and invoke an appropriate Struts Action. Each Action is configured to perform a specific function in the application, such as add an image to a compilation, recommend an image from one user to another, add a print of an image to a user's shopping cart, etc.
When a Struts Action completes its work, which may be assisted by an application-specific business logic layer—the Action invokes a Java Server Page 136 (or other interface component) to deliver presentation markup code to the user. This code may include references to any number of images.
Business Logic 138 comprises application-specific logic for performing back-end tasks such as resizing an image (e.g., as a thumbnail, for full screen viewing), issuing an electronic mail message to a user to inform her that new images have been added to her community, integrating prints and orders for photo books with a vendor's order processing system, etc.
Image application server 130 may comprise a java virtual machine or similar environment for supporting object oriented programming using the Java programming language and/or other suitable programming languages.
A central server includes any number, type and configuration of storage devices 122 for storing images. Additionally, some images manipulated or accessed by central server 120 may be stored on users' devices. In particular, a user may create a new image and temporarily or permanently store it on his or her local device. The system may also upload a copy to central server 120 for purposes of sharing with other users, generating an image compilation and so on. In one implementation, image stores 122 may store millions of images and thus comprise storage capacity in excess of one petabyte. A database or data server for managing access to the stored images may be included in central server 120 (e.g., as part of image application server 130 or image stores 122).
User interaction with central server 120 and manipulation of images can be logically separated into four layers. A presentation layer controls the presentation of images to a user on his or her computing/communication device. Logic for the presentation layer may be designed using HTML (HyperText Markup Language), Adobe Flash (e.g., for displaying or laying out images) and/or other suitable code or models. Illustratively, presentation logic may be stored on a central server (e.g., with web server 124 or some component not depicted in
Alternatively, however, presentation logic may be stored on a user's device as part of a native application program, and therefore may not need to download presentation layer logic from a central server. In this alternative implementation, the collaborative sharing of images may operate in a peer-to-peer mode instead of (or in addition to) a centralized mode. In particular, storage of images may occur only (or partially) on user devices.
A communication layer handles communication between central server 120 and presentation layer logic operating on a user's computing/communication device. The communication layer may deliver the presentation layer logic as needed (e.g., when the user connects to the central server), and also delivers user requests (e.g., for images) to the central server processing logic, and conveys any response.
The processing logic layer sequences user activity within the system and services user requests, but also manages and presents images as needed. The processing logic layer may manage user sessions and deliver appropriate interfaces or options (e.g., web pages, application pages, java server pages) to a user based on the appropriate options for the user's current state. Thus, the processing logic may be required to determine what images should be delivered to a user. The processing logic also receives data from users—not only new images, but also rankings, recommendations, descriptions and/or other metadata regarding images.
Finally, a data management layer of logic manages the storage, retrieval and caching of information about users, communities and images. All persistent information relating to user interaction with images, image metadata, the organization of images and other information about the state of the image collaboration system is handled by this layer. The data management layer may comprise a suitable database management system, such as a MySQL database.
In different embodiments of the invention, different criteria may be employed to rate/rank an image or estimate its interest to a specific user or a group of users. Such criteria include information regarding the image, such as the subject, an event or setting in which it was captured, image quality, date and/or time, size, type (e.g., bitmap, jpeg), clarity, etc. Criteria for estimating an image's interest may also include activity involving the image, such as whether (or how often) it was recommended or forwarded from one user to another, whether it was included in a compilation of images, how often it was accessed (e.g., downloaded, viewed), whether it was annotated by a user (e.g., to add a description or caption), and so on. Yet other criteria may also be considered, such as relationships between users making and receiving a recommendation regarding the image, the community (or communities) to which a user who receives a recommendation belongs to, other users in the same community as a user receiving a recommendation, etc.
The following is an illustrative set of metrics that may be employed in an embodiment of the invention to determine a user's estimated level of interest in an image. These metrics need not all be employed, and do no limit the criteria that may be employed in other embodiments.
Recommendation: A community user may recommend an image to another community user. An image P recommended by user A to user B will be ranked higher for user B (regarding estimated interest) than an image not recommended to user B. The weighting given to this criterion may increase in proportion to the number of recommendations issued to user B.
Similarly, the more recommendations issued by user A regarding image P, the higher the ranking assigned to image P for user A.
Recommendation from Relative: If user A recommends image P1 to user B and they are related (e.g., by blood or interaction), and user C recommends image P2 to user B but they are not related, image P1 will be rated higher than image P2 for user B.
The weighting given to a recommendation between related users may vary with the level of interaction (e.g., frequent communications) or closeness of relation (e.g., husband and wife) of the users. Two users may be considered “related” if they appear in the same image. Their level of “relatedness” may increase according to the frequency with which they appear together in different images, so that their recommendations to each other will carry more and more weight. Identities of users appearing in an image may be included in the metadata regarding an image.
Two users may also be considered “related” if they have similar interest in an image. For example, if user A recommends image P1 to user B and user B includes the image in a compilation or a set of favorites, but did not include an image P2 recommended by user C, a later recommendation from user A to user B regarding image P3 will carry more weight than a recommendation from user C to user B regarding some other image. The weight may increase with the number of recommendations.
Recommendation between Other Users: A user's estimated interest in an image may also be affected by recommendations between other users, especially if the other users are in the same community. Thus, if user B views an image that was not recommended to her specifically, but was recommended by user A to user C, that image may be ranked higher for user B than an image not recommended by another user. The ranking may increase with the number of recommendations.
Recommendation Exclusivity: An image P1 recommended just to user B may be ranked higher for user B than an image P2 recommended to multiple users including user B. Illustratively, image P1 may be a portrait of user B while image P2 is a group photo.
Image Creator: Images created or processed by one user may be consistently more appealing to users than images provided by someone else. Thus, if user A creates image P1 and image P1 is recommended more frequently (e.g., within the same community) than an image P2 created by user C, a later image created by user A may be ranked higher than a later image created by user C. Creation of an image may involve taking a digital photograph, generating a digital image via graphics or drawing software, etc.
Explicit Rating: A user may explicitly rate or rank an image. Images with higher ratings may then be weighted more heavily (regarding other user's likely interest) than images with lower ratings. As described above regarding recommendations, an explicit rating from user A may be given different weights when determining how interesting an image is for user B, depending on whether and how user A and user B are related. Thus, if user A and user B are members of the same family, their ratings may be weighted more heavily for each other than ratings from other users.
Compilation Exclusivity: An image added to a compilation (e.g., electronic album, set of favorites) just by user B may be ranked higher for user B than an image added to compilations by user B and by other users.
In addition, however, an image added to a compilation of user B may be ranked higher for a different user in user B's community than an image not included in any community user's compilations.
Image Resolution: A larger image or an image of relatively high resolution may be ranked higher than an image of lower resolution or size.
Image Clarity: Sharper images may be ranked higher than unclear images. Clarity of a digital image may be measured by a standard image processing technique, such as by measuring the occurrence of high frequencies in fast Fourier transforms.
User Interaction with Images: As users view images via the image collaboration system (e.g., images served from image application server 130 of
For example, as a user views a relatively large version of an image (e.g., expanded from a thumbnail by hovering over the thumbnail), a ranking for that image may be increased. If a user has to click on a thumbnail, icon or description of an image to access a larger view, the ranking of that image may increase. The ranking of an image may also increase with the number of times the image is viewed or otherwise accessed. Yet further, the length of time between a user's access of an image and a subsequent access of the same image by the same user may increase or decrease the image's ranking.
Metadata: The ranking of an image for a user may increase when that user (or possibly a related user) submits metadata for the image, such as a description, caption, event, subject(s), etc.
Subject Exclusivity: An image's ranking may be affected by the number of subjects identified for an image (e.g., the more people that appear in the image). For example, if user B is identified as the only subject in image P1, and just one of multiple subjects in image P2, image P1 may be ranked higher for user B than image P2.
As can be seen by the variety of illustrative metrics described above, information for estimating interest in an image may be gathered directly from users (e.g., as an explicit rating), may be inferred from a user's activity (e.g., whether or how often she accesses an image), may be inferred from other user's activities, etc. Activities of a user may be more relevant in determining the interest of a “related” user, wherein one user may be considered related to another if they are kin or if they interact relatively frequently (e.g., exchanging images and/or recommendations, reviewing images created by each other).
For any or all of the metrics, individual values may be derived for any tuple {user, image}. Thus, separate values may be generated for each user, for each image. The values may be weighted according to how closely the corresponding metric is believed to correlate with user interest. Default weights may be used, and/or context-specific weights may be applied.
Illustratively, one set of weights may be applied for the purpose of ordering images to be presented a user who wishes to view images she has not yet seen. A different set of weights may be employed to order or arrange images for printing in a physical compilation. For example, image resolution may be critical in the latter situation, but almost irrelevant in the former.
Weights for modifying metric values may be set by a system administrator (e.g., an operator of image application server 130 of
Weights may be modified based on the composition of a user community. For example, if all or most users in a given community are related, then metrics affected by users' relationships may not be very useful. Conversely, if another community is very large and has relatively few inter-related users, then relationships may be very important.
Individual users may help tune weights employed for estimating image interest for those users. Illustratively, the user could answer a set of questions designed to help formulate optimal weighting.
Weights may also, or instead, be adaptive or dynamic. For example, the accuracy of a given set of weights W1 for user B could be observed by examining how often user B selects resulting highly ranked images as favorites, includes them in compilations, recommends them to family members or other related users, etc. If the correlation is relatively weak, the system may adopt a different set of weights W2 for user B and determine whether set W2 works better, and so on.
Images need not always be presented or arranged for a user in decreasing order of their estimated interest for that user. Some or all may be presented in reverse order (e.g., to provide incentive for continued viewing) or in a random order.
When a user views or otherwise accesses an image (e.g., reviews a list of image captions or titles, scans a selection of thumbnails), he may be asked or prompted to recommend the image to one or more other users, forward a copy (e.g., via electronic mail), download it, save it as wallpaper or a background on his computing device, request a hard-copy (e.g., physical print, photo book, album), add it to a compilation, submit metadata regarding the image (e.g., description, event, subject), rate it according to interest or quality, and so on.
In one embodiment of the invention, some form of intelligent subject determination may be applied to an image. The image collaboration system can therefore facilitate users' searches for images containing particular people or other subjects.
Illustratively, a subject of an image may be inferred from a user recommendation. For example, if image P is recommended to user B, there is some probability that user B (or at least some subject of interest to user B) appears in image P. The probability may increase proportionately to the number of times image P is recommended to user B by different users, especially if it is recommended to few (or no) other users. An image that is recommended to many different users, including user B, may include user B as a subject but is likely to be a group image.
A subject of an image may also be inferred by examining which user or users add the image to their favorites or some other compilation of images. In particular, an image selected by user B as the lead, first or other principal image of a compilation may be likely to include that user as a subject. Similarly, the title of a compilation that includes a particular image, and/or metadata of the image (e.g., caption, description) may give a clue as to a subject of the image.
In another embodiment of the invention, explicit or learned relationships between users may be factors in selecting images for presentation to a user or in weighting metrics used to facilitate image selection. For example, if users A and B appear together (e.g., are subjects) within a given collection of images, some relationship may be assumed to exist between them. Then, when user B requests a compilation of images (e.g., a slideshow, an album, a photo book), images may be drawn from that collection because of the users' relationship. Depending on the strength of user B's relationship with user A (e.g., how many images they appear in together), more images for the compilation may be selected from that collection than from a different collection reflecting user B's relationship with user C.
In an illustrative embodiment of the invention, names, captions or titles for images or for a set of images reflecting a particular event may be derived based on information provided by the image creators and/or information gleaned from the users' client devices. For example, digital photographs of a school's Tuesday afternoon soccer match taken by different users may be correlated based on their date/time of creation or upload, the fact that the users are members of a user community associated with the school, etc. The image collaboration system can then automatically suggest an appropriate name for the collection of images.
An image collaboration system may support multiple versions of a single image. Illustratively, any user within a community may edit any image within that community unless it is restricted. An edited image may be stored as a new version of the image, as an entirely new image, or may even replace the original image (in one implementation only the creator of an image or an administrator may replace an image or delete an image from the community). An edited version of an image may alternatively be kept private for the editor and any user the editor wishes to share it with. The extent, timing, frequency and/or other characteristics of edits to an image may affect its ranking.
If an image is replaced with a new version, it may or may not automatically be replaced in every electronic compilation in which it appears. For example, users may be asked if they wish to replace the version(s) in their compilations. Automatic replacement may depend on the type of modification made to the original image, whether a relationship exists between the editor and a user who has the image in a compilation, and/or other factors.
In an embodiment of the invention, any set of images may be shared among all users in a given community, or may be restricted to some subset of users within the community or even to a single user. In this embodiment, the creator of an image or an administrator may restrict access to the image as desired.
From a set of images available to a user, the user may select any subset to form a compilation “owned” by that user. The images may be drawn from a general community pool of images and/or from other images the user is able to access—such as images within other users' compilations. Thus, the user may adopt a compilation assembled by some other user and then modify it, with or without affecting the other user's compilation, depending on the user's access rights. A compilation may include sub-compilations.
The image collaboration system may assist or facilitate a user's selection of images for a compilation, or may select a set of images for the user and allow the user to add/delete images to/from those selected. For example, a user may have the system display the N images estimated to be of most interest to that user. Or, the user may ask for the N most popular images from a given set (e.g., all images relating to a particular event), images that some other particular user (e.g., a family member) found most interesting, etc. The system can also perform automatic layout and/or facilitate a user's layout of a compilation.
A compilation assembled by a user (with or without selections made by the system) may be retained electronically, for viewing on the user's computing or communication device, or may be used to generate a physical compilation (e.g., a photo book, a yearbook, a catalog).
Thus, in an embodiment of the invention, personalized compilations (electronic and/or physical) can be easily assembled. A “general” compilation may be organized by the system or a first user for an entire community, and may have approximately equal interest to all community members. From that general compilation, personalized or tailored compilations may be derived that include images that are more interesting to a specific user or limited group of users (e.g., a family).
For example, a school may assemble a general compilation of some interest to most or all students' families. Tailored compilations may then be developed for individual grades, classes, teams, clubs, students and so on. A personalized compilation may therefore focus on a particular student, her friends, her activities, her classes, etc.
Images may also be shared across communities. In one embodiment, a user who is a member of two communities may set a policy specifying that images added to one community are automatically added to the other if they match some specified criteria, such as subject(s), event, location, etc. For example, two communities corresponding to two different schools or organizations could share images that feature students or members of both. Similarly, an image could be restricted from being shared between communities (or users within a community) based on subject, event or other criteria.
In one embodiment of the invention, an image collaboration system may have relationships with people and other entities outside its communities and community members, to enhance the collaboration or make additional services available, or for other purposes. For example, the system may interface with photographers, artists, photo retouching services, photo organizing services, and so on. These services may be allowed to advertise via the system and users may be able to contract for the services via the system.
The system may provide anonymity to users, so that when an image is provided to an outside service, names subjects of the image (e.g., people in the image) may be replaced with other identifiers or fictitious names. Thus, a user may select or have the system select a set of images for printing as a photo book, and the images may be provided to a suitable service provider without divulging private information.
In operation 202, one or more user communities are established. A community may be associated with an institution or organization, such as a school, a church, a neighborhood association, a team. Or a community may be based on some other interest or activity common to its members, such as a sport, a hobby, and so on. A community may not even have an associated theme, and may simply provide means for sharing and collaborating on images.
Also in operation 202, images are deposited with a community. It may be typical for the number of images to far exceed the number of community members. For example, a community may have tens, hundreds or thousands of members, but may encompass thousands, tens of thousands or hundreds of thousands of images.
A community may be established with any number of member users, and may include sub-communities or groups. Such sub-communities may be defined by any member (or an administrator), and may facilitate restricted sharing of some images. Unless restricted, an image created or submitted within a community may automatically be made available to all members of the community.
In operation 204, a user (“user A”) accesses some or all of the community images to select some favorites and make them available for easy retrieval, select some to place into a compilation (e.g., an electronic album of pictures from a school event), download some images, view some images, create or submit an image and/or take other action regarding images. For example, she may rate an image based on how much she enjoys them, whether she thinks the image is good (e.g., clear, in focus, not blurry), whether she thinks it should be removed from the community, etc. She may also create or submit any number of new images (e.g., new photos uploaded from her digital camera).
In operation 206, user A recommends one or more images to another user. The other user may be a family member, a friend or even someone user A barely knows. For example, while reviewing some community images user A may find image P, a particularly good photo of a child of user B, and therefore issue a recommendation of image P to user B.
In operation 208 (and throughout operation of the image collaboration system), the system accumulates data regarding user activity. For example, it tracks image recommendations, accesses (e.g., views, downloads, copies, forwards), inclusions in compilations, explicit (and implicit) ratings or rankings, etc. The system may also note relationships between users, which may be explicit (e.g., user A's profile may indicate that user B is a relative) or implicit (e.g., user A may recommend images to user B relatively frequently).
In operation 210, user B views the image recommended by user A, and/or other images, and decides he wants a compilation of images, which may be electronic or physical. He therefore prompts the system to assemble a compilation of some default number (e.g., twenty, forty) of images, or may specify any number of criteria for the compilation (e.g., number of images, minimum or maximum image size, minimum resolution, event to which the images correspond).
In operation 212, unless any criteria specified by user B conflict, the system automatically selects a number of images that have been estimated to be most interesting to user B. As described above, many different metrics may be considered by the system in rating or ranking the images. In one implementation of this embodiment of the invention, the system applies some or all of the metrics listed in TABLE 1, which provides a description of each metric and a default weight. As described below, values for the various metrics may be modified according to the weights, with the results being manipulated to yield images' rankings.
In operation 214, values for any number of the metrics (e.g., metrics marked with an “*” in TABLE 1) may be normalized to values between zero and one (inclusive or exclusive). In one implementation an exponential function is used to produce the normal values. For example, a normalized value Mi for the ith metric may be calculated as:
Mi=1−D*e(−x)
wherein D may be a constant (e.g., 1) or may be dynamic (in order to enable further variability in the calculation of image ratings), and x is the value of the ith metric prior to normalization.
In operation 216, the interest rating or ranking of an image “R” can then be calculated from its associated (normalized) metric values and weights as follows, wherein Wi is the weight assigned to metric Mi:
R=Σ(Mi*Wi)/Σ(Wi)
Ratings may be calculated for any number of images, such as all images in the community or all images meeting certain criteria (e.g., date/time, event, subject). Depending on the type or nature of the requested compilation, metric weights may be adjusted accordingly. Thus, a metric reflecting how many other users requested a physical photo book including image P may be weighted higher if user B has requested a physical photo book than if he requested an electronic album.
In operation 218, the rated images are sorted based on their ratings.
In operation 220 in this embodiment of the invention, the top N images are offered to user B as suggested images for the requested compilation. User B may add images to the compilation, remove images, alter the selection criteria and request a new compilation, etc.
In the illustrated embodiment of the invention, metric values may be calculated each time they are needed (e.g., when a user requests generation of a suggested set of images). Alternatively, metric values may be continually updated based on user activity. Thus, each time an image P1 is recommended to user B, the NumTotalRecommendations and NumRecommendationsToSelf metrics (see TABLE 1) may be incremented for the tuple {user B, P1}. Metric values may therefore be cached (e.g., on an image application server) and updated as needed.
Methods of rating images described herein may be used to select sets of images for compilations, but may also be used to sort images for presentation to a user for other purposes. For example, a user may connect to the image collaboration system and wish to view images she has not yet seen (e.g., images added to the system since her last visit). The system may rate the unseen images (or retrieve such ratings) to determine an order in which they should be presented to her.
The program environment in which a present embodiment of the invention is executed illustratively incorporates a general-purpose computer or a special purpose device such as a hand-held computer. Details of such devices (e.g., processor, memory, data storage, display) may be omitted for the sake of clarity.
It should also be understood that the techniques of the present invention may be implemented using a variety of technologies. For example, the methods described herein may be implemented in software executing on a computer system, or implemented in hardware utilizing either a combination of microprocessors or other specially designed application specific integrated circuits, programmable logic devices, or various combinations thereof. In particular, the methods described herein may be implemented by a series of computer-executable instructions residing on a suitable computer-readable medium. Suitable computer-readable media may include volatile (e.g., RAM) and/or non-volatile (e.g., ROM, disk, DVD) memory, carrier waves and transmission media (e.g., copper wire, coaxial cable, fiber optic media). Exemplary carrier waves may take the form of electrical, electromagnetic or optical signals conveying digital data streams along a local network, a publicly accessible network such as the Internet or some other communication link.
The foregoing embodiments of the invention have been presented for purposes of illustration and description only. They are not intended to be exhaustive or to limit the invention to the forms disclosed. Accordingly, the scope of the invention is defined by the appended claims, not the preceding disclosure.
Claims
1. A method of selecting images for a personalized compilation of images, the method comprising:
- maintaining a pool of images for a community of users;
- weighting each of the images based on characteristics of said image, including: image quality; and image clarity;
- further weighting each of the images for each user based on said user's interaction with said image;
- inferring a relationship between multiple users in the community of users based on said users' interactions with a subset of images in the pool of images;
- adjusting said image weightings for each of the multiple users based on the related users' interaction with the subset of images; and
- selecting images for a personalized compilation for a given user of the community of users based on said weightings.
2. The method of claim 1, wherein a user's interaction with an image includes one or more of:
- a duration of time spent viewing the image;
- a frequency with which the image is viewed; and
- a recommendation of the image to another user.
3. The method of claim 1, further comprising:
- further weighting an image for a user based on the exclusivity of said user's interaction with said image, wherein the user's interaction with an image is exclusive if no other user in the community of users has the same interaction with said image.
4. The method of claim 1, further comprising:
- further weighting an image for a user based on the exclusivity of said user's interaction with said image, wherein said exclusivity comprises viewing a version of the image not viewed by other users in the community.
5. The method of claim 1, further comprising:
- further weighting an image for a user based on the exclusivity of said user's interaction with said image, wherein said exclusivity comprises receiving one of a limited number of recommendations of the image from another user within the community.
6. The method of claim 1, wherein said inferring comprises:
- determining that the multiple users had identical interaction with the subset of images.
7. The method of claim 1, wherein said inferring comprises:
- determining that the multiple users had similar interaction with the subset of images.
8. The method of claim 7, wherein said similar interaction comprises one or more of:
- viewing the subset of images;
- printing the subset of images;
- receiving a recommendation regarding the subset of images; and
- corresponding regarding the subset of images.
9. The method of claim 1, further comprising:
- automatically determining a subject of a first image in the pool; and
- weighting the first image based on weights associated with other images containing the subject.
10. A computer readable medium storing instructions that, when executed by a computer, cause the computer to perform a method of selecting images for a personalized compilation of images, the method comprising:
- maintaining a pool of images for a community of users;
- weighting each of the images based on characteristics of said image, including: image quality; and image clarity;
- further weighting each of the images for each user based on said user's interaction with said image;
- inferring a relationship between multiple users in the community of users based on said users' interactions with a subset of images in the pool of images;
- adjusting said image weightings for each of the multiple users based on the related users' interaction with the subset of images; and
- selecting images for a personalized compilation for a given user of the community of users based on said weightings.
11. A computerized system for compiling personalized sets of images for different users, comprising:
- storage means for maintaining a pool of images available for access by a community of users;
- recording means for recording activity regarding the images, including: a duration of time a user views an image; a frequency with which a user views an image; addition of a caption to an image; requesting an enlarged view of a viewed image; and recommendation of an image to a single user by multiple other users;
- means for determining possible subjects of one or more images by noting: which users access said image; which users, if any, said image is recommended to by the users who access said image; and any relationships between the users who access said image and users to whom they recommend said image; and
- means for automatically compiling personalized sets of images likely to be of interest to the different users.
Type: Application
Filed: Oct 31, 2007
Publication Date: Apr 3, 2008
Applicant: Picateers, Inc. (San Mateo, CA)
Inventors: Lawrence Jacobs (Redwood City, CA), Shehzaad Nakhoda (Mountain View, CA), Xiang Liu (Fremont, CA), Tie Zhong (San Jose, CA)
Application Number: 11/982,021
International Classification: G06K 9/68 (20060101);
