# US 2023/0410508 A1 · Systems, methods, and apparatuses for selecting/generating visual representations of media content > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2023/0410508 A1 - Document ID: US20230410508A1 - Status: Published application - Kind: A1 - Application number: 17/843,250 - Filed: 2022-06-17 - Published: 2023-12-21 - Earliest claimed: 2022-06-17 - Assignees: Comcast Cable Communications, LLC - Applicants: Comcast Cable Communications, LLC - Inventors: Ehsan Younessian - Web page: https://patentgrep.com/patent/US20230410508A1 ## CPC - G06V 20/46: {Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames} - G06V 10/764: using classification, e.g. of video objects - G06V 20/41: {Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items (segmenting video sequences G06V20/49)} - H04N 21/44008: {involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream (arrangements characterised by components specially adapted for monitoring, identification or recognition of video in broadcast systems H04H60/59)} - H04N 21/4532: {involving end-user characteristics, e.g. viewer profile, preferences (monitoring of user activities for profile generation for accessing a video database G06F16/739; user profiles in network data switching protocols H04L67/306; processing of user preferences or user profiles in wireless networks H04W8/18)} - H04N 21/482: End-user interface for programme selection - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION ## Abstract Content items may be represented in a listing by one or more thumbnails containing images from the content items. The images of a content item may be processed to recognize persons and/or objects, and frames of the content item may be selected based on interest values of the recognized persons and/or objects. For example, frames may be selected based on having uncommon (rare) objects and/or common (popular) persons. ## Claims ### Claim 1 A method comprising: performing an image recognition process on frames of a content item; selecting, based on rarity of an object associated with an image recognized by the image recognition process, a frame of the content item; generating a visual representation of the content item, wherein the visual representation is based on the selected frame; and causing output of the visual representation of the content item. ### Claim 2 (depends on 1) The method according to claim 1, wherein the selecting the frame of the content item comprises assigning different interest values to different types of people and/or objects recognized in the selected frame. ### Claim 3 (depends on 1) The method according to claim 1, wherein the selecting the frame of the content item comprises: assigning a first interest value, to a first type of recognized image, wherein the first interest value is set higher to higher quantities of times that the first type of recognized image is recognized in the content item; and assigning a second interest value, to a second type of recognized image, wherein the second interest value is set lower to higher quantities of times that the second type of recognized image is recognized in the content item. ### Claim 4 (depends on 1) The method according to claim 1, wherein the selecting the frame of the content item is based on: a first value assigned to a first type of recognized image recognized in the content item; and a second value assigned to a second type of recognized image recognized in the content item. ### Claim 5 (depends on 1) The method according to claim 1, wherein the selecting the frame of the content item is based on: a first value associated with a person in the content item; and a second value associated with an object in the content item. ### Claim 6 (depends on 1) The method according to claim 1, wherein the rarity is based on a genre of the content item. ### Claim 7 (depends on 1) The method according to claim 1, wherein the selecting the frame of the content item is further based on a user profile. ### Claim 8 A method comprising: receiving image streams of a plurality of content items, wherein each image stream comprises a plurality of video frames for a corresponding content item of the plurality of content items; selecting, for each of the plurality of content items, based on rarity associated with objects recognized in the pluralities of video frames, a corresponding video frame to represent the corresponding content item; and causing output of a visual listing of the plurality of content items, wherein each of the plurality of content items is represented, in the visual listing, by the corresponding video frame selected for the content item. ### Claim 9 (depends on 8) The method according to claim 8, further comprising assigning different values to different types of images recognized in the pluralities of video frames. ### Claim 10 (depends on 8) The method according to claim 8, further comprising generating information indicating different interest values for different people and objects recognized in the pluralities of video frames by: assigning a first value to people in the pluralities of video frames, wherein the first value is set higher to higher quantities of times that people are recognized in the pluralities of video frames; and assigning a second value to objects in the pluralities of video frames, wherein the second value is set lower to higher quantities of times that objects are recognized in the pluralities of video frames. ### Claim 11 (depends on 8) The method according to claim 8, wherein the selecting the corresponding video frame comprises: assigning a first value to a person recognized in the pluralities of video frames; and assigning a second value to an object recognized in the pluralities of video frames. ### Claim 12 (depends on 8) The method according to claim 8, wherein the selecting the corresponding video frame comprises combining: first interest values associated with people recognized in the pluralities of video frames; and second interest values associated with objects recognized in the pluralities of video frames. ### Claim 13 (depends on 8) The method according to claim 8, wherein the selecting the corresponding video frame is further based on genres of content items containing the pluralities of video frames. ### Claim 14 (depends on 8) The method according to claim 8, wherein the selecting the corresponding video frame is further based on a user profile. ### Claim 15 A method comprising: receiving, by a computing device, information indicating a plurality of persons or objects, and for each of the persons or objects, a corresponding quantity of times that the person or object was recognized in a library of content items; based on the information and based on rarity associated with objects recognized in the library of content items, selecting a frame from a content item; using the selected frame to generate an image to represent the content item in a content item listing; and causing, based on selection of the content item from the content item listing, output of the selected content item. ### Claim 16 (depends on 15) The method according to claim 15, further comprising assigning interest values to persons and objects recognized in frames of the content item, and wherein: higher interest values are assigned to persons that are recognized more frequently in the library of content items; and lower interest values are assigned to objects that are recognized more frequently in the library of content items. ### Claim 17 (depends on 15) The method according to claim 15, further comprising assigning interest values to persons and objects recognized in frames of the content item, and wherein the interest values are weighted based on a genre of the content item. ### Claim 18 (depends on 15) The method according to claim 15, further comprising assigning interest values to persons and objects recognized in frames of the content item, and wherein the interest values are weighted based on a user profile. ### Claim 19 (depends on 15) The method according to claim 15, wherein the selecting the frame is also based on combining interest values of: a person recognized in the frame; and an object recognized in the frame. ### Claim 20 (depends on 15) The method according to claim 15, further comprising: performing an image recognition process to recognize one or more persons or objects in the selected frame; and sending the selected content item to a user device based on the selection of the content item from the content item listing. ### Claim 21 (depends on 1) The method according to claim 1, wherein: the selecting comprises selecting, based on determining that a first object in the content item is rarer than a second object in the content, the frame, of the content item, comprising the first object; and the visual representation comprises the selected frame. ### Claim 22 (depends on 1) The method according to claim 1, wherein the content item is an episode of a program series, and the rarity of the object is based on a quantity of times the object appears in the program series. ### Claim 23 (depends on 8) The method according to claim 8, wherein the selecting comprises selecting, based on a first object, in the corresponding content item, being rarer than a second object in the corresponding content item, the corresponding video frame comprising the first object. ### Claim 24 (depends on 8) The method according to claim 8, wherein the visual listing comprises a listing of selectable video programs, and each of the video programs is represented by a thumbnail image of a scene from the video program. ### Claim 25 (depends on 15) The method according to claim 15, further comprising determining that: a first frame, of a particular content item, comprises a first object; and a second frame, of the particular content item, comprises a second object, wherein the selecting is based on determining that the first object was recognized, in the library, fewer times than the second object was recognized in the library.