# US 11,113,575 B2 · Automatic image selection for online product catalogs > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,113,575 B2 - Document ID: US11113575B2 - Status: Granted - Kind: B2 - Application number: 16/566,121 - Filed: 2019-09-10 - Published: 2021-03-11 - Granted: 2021-09-07 - Earliest claimed: 2019-09-10 - Assignees: eBay Inc. - Applicants: eBay Inc. - Inventors: Arnon Dagan; Ido Guy; Alexander Nus; Raphael Bryl; Noa Shimoni Barzilai; Avinoam Omer; Yan Radovilsky; Einav Itamar; Gadi Mikles - Web page: https://patentgrep.com/patent/US11113575B2 ## CPC - G06N 20/00: Machine learning - G06F 18/2148: {characterised by the process organisation or structure, e.g. boosting cascade} - G06F 18/2178: {based on feedback of a supervisor} - G06F 18/2321: using statistics or function optimisation, e.g. modelling of probability density functions - G06F 18/24155: {Bayesian classification} - G06N 3/0464: Convolutional networks [CNN, ConvNet] - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06V 10/764: using classification, e.g. of video objects - G06V 10/7784: {based on feedback from supervisors} - G06V 20/35: {Categorising the entire scene, e.g. birthday party or wedding scene} - G06V 20/70: Labelling scene content, e.g. deriving syntactic or semantic representations - G06N 3/045: Combinations of networks - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract Disclosed are systems, methods, and non-transitory computer-readable media for automatic image selection for online product catalogs. An image selection system gathers feature data for images of an item included in listings posted to an online marketplace. The image selection system uses the feature data as input in a machine learning model to determine probability scores indicating an estimated probability that each image is suitable to represent the item. The machine learning model is trained based on a set of training images of the item that have been labeled to indicate whether they are suitable to represent the image. The image selection system compares the probability scores and selects an image to represent the item as a stock image based on the comparison. ## Claims ### Claim 1 A method comprising: gathering, a first set of feature data for a first image included as part of a first listing for a first item, the first set of feature data including image features describing the first image and user features describing a first user that posted the first listing for the first item; determining a first probability score for the first image by using the first set of feature data as input in a machine learning model, the first probability score indicating an estimated probability that the first image is suitable to represent the first item, the machine learning model having been trained based on a set of training images of the first item and sets of feature data associated with the set of training images; gathering, a second set of feature data for a second image included as part of a second listing for the first item, the second set of feature data including image features describing the second image and user features describing a second user that posted the second listing for the first item; determining a second probability score for the second image by using the second set of feature data as input in the machine learning model, the second probability score indicating an estimated probability that the second image is suitable to represent the first item; and selecting, based on a comparison of at least the first probability score and the second probability score, the first image to represent the first item. ### Claim 2 (depends on 1) The method of claim 1, wherein selecting the first image to represent the first item comprises: determining, based on the comparison of at least the first probability score and the second probability score, that the first probability score is greater than the second probability score. ### Claim 3 (depends on 1) The method of claim 1, wherein the first set of feature data for the first image also includes item features describing the first item. ### Claim 4 (depends on 1) The method of claim 1, wherein the first set of feature data for the first image also includes categorical features describing a category of items in which the first item is included. ### Claim 5 (depends on 1) The method of claim 1, wherein each training image from the set of training images is labeled to indicate whether the respective training image is suitable to represent the first item. ### Claim 6 (depends on 1) The method of claim 1, further comprising: gathering, a third set of feature data for a third image included as part of a third listing for the first item, the third set of feature data including image features describing the third image and user features describing a third user that posted the third listing for the first item; and determining a third probability score for the third image by using the third set of feature data as input in the machine learning model, the third probability score indicating on estimated probability that the third image is suitable to represent the first item, wherein the comparison of at least the first probability score and the second probability score also includes the third probability score. ### Claim 7 (depends on 1) The method of claim 1, wherein the first image is selected to represent the first item by representing a category of items that encompasses the first item. ### Claim 8 A system comprising: one or more computer processors; and one or more computer-readable mediums storing instructions that, when executed by the one or more computer processors, cause the system to perform operations comprising: gathering, a first set of feature data for a first image included as part of a first listing for a first item, the first set of feature data including image features describing the first image and user features describing a first user that posted the first listing for the first item; determining a first probability score for the first image by using the first set of feature data as input in a machine learning model, the first probability score indicating an estimated probability that the first image is suitable to represent the first item, the machine learning model having been trained based on a set of training images of the first item and sets of feature data associated with the set of training images; gathering, a second set of feature data for a second image included as part of a second listing for the first item, the second set of feature data including image features describing the second image and user features describing a second user that posted the second listing for the first item; determining a second probability score for the second image by using the second set of feature data as input in the machine learning model, the second probability score indicating an estimated probability that the second image is suitable to represent the first item; and selecting, based on a comparison of at least the first probability score and the second probability score, the first image to represent the first item. ### Claim 9 (depends on 8) The system of claim 8, wherein selecting the first image to represent the first item comprises: determining, based on the comparison of at least the first probability score and the second probability score, that the first probability score is greater than the second probability score. ### Claim 10 (depends on 8) The system of claim 8, wherein the first set of feature data for the first image also includes item features describing the first item. ### Claim 11 (depends on 8) The system of claim 8, wherein the first set of feature data for the first image also includes categorical features describing a category of items in which the first item is included. ### Claim 12 (depends on 8) The system of claim 8, wherein each training image from the set of training images is labeled to indicate whether the respective training image is suitable to represent the first item. ### Claim 13 (depends on 8) The system of claim 8, the operations further comprising: gathering, a third set of feature data for a third image included as part of a third listing for the first item, the third set of feature data including image features describing the third image and user features describing a third user that posted the third listing for the first item; and determining a third probability score for the third image by using the third set of feature data as input in the machine learning model, the third probability score indicating on estimated probability that the third image is suitable to represent the first item, wherein the comparison of at least the first probability score and the second probability score also includes the third probability score. ### Claim 14 (depends on 8) The system of claim 8, wherein the first image is selected to represent the first item by representing a category of items that encompasses the first item. ### Claim 15 A non-transitory computer-readable medium storing instructions that, when executed by one or more computer processors of one or more computing devices, cause the one or more computing devices to perform operations comprising: gathering, a first set of feature data for a first image included as part of a first listing for a first item, the first set of feature data including image features describing the first image and user features describing a first user that posted the first listing for the first item; determining a first probability score for the first image by using the first set of feature data as input in a machine learning model, the first probability score indicating an estimated probability that the first image is suitable to represent the first item, the machine learning model having been trained based on a set of training images of the first item and sets of feature data associated with the set of training images; gathering, a second set of feature data for a second image included as part of a second listing for the first item, the second set of feature data including image features describing the second image and user features describing a second user that posted the second listing for the first item; determining a second probability score for the second image by using the second set of feature data as input in the machine learning model, the second probability score indicating an estimated probability that the second image is suitable to represent the first item; and selecting, based on a comparison of at least the first probability score and the second probability score, the first image to represent the first item. ### Claim 16 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein selecting the first image to represent the first item comprises: determining, based on the comparison of at least the first probability score and the second probability score, that the first probability score is greater than the second probability score. ### Claim 17 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the first set of feature data for the first image also includes item features describing the first item. ### Claim 18 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein the first set of feature data for the first image also includes categorical features describing a category of items in which the first item is included. ### Claim 19 (depends on 15) The non-transitory computer-readable medium of claim 15, wherein each training image from the set of training images is labeled to indicate whether the respective training image is suitable to represent the first item. ### Claim 20 (depends on 15) The non-transitory computer-readable medium of claim 15, the operations further comprising: gathering, a third set of feature data for a third image included as part of a third listing for the first item, the third set of feature data including image features describing the third image and user features describing a third user that posted the third listing for the first item; and determining a third probability score for the third image by using the third set of feature data as input in the machine learning model, the third probability score indicating on estimated probability that the third image is suitable to represent the first item, wherein the comparison of at least the first probability score and the second probability score also includes the third probability score.