# US 10,872,232 B2 · Image filtering for subject and crowd identification > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 10,872,232 B2 - Document ID: US10872232B2 - Status: Granted - Kind: B2 - Application number: 16/195,580 - Filed: 2018-11-19 - Published: 2020-05-21 - Granted: 2020-12-22 - Earliest claimed: 2018-11-19 - Assignees: Adobe Inc. - Applicants: Adobe Inc. - Inventors: Subham Gupta; Ajay Bedi - Web page: https://patentgrep.com/patent/US10872232B2 ## CPC - G06V 40/161: {Detection; Localisation; Normalisation} - G06F 18/23213: with fixed number of clusters, e.g. K-means clustering - G06V 10/763: {Non-hierarchical techniques, e.g. based on statistics of modelling distributions} - G06V 40/168: {Feature extraction; Face representation} - G06V 40/172: {Classification, e.g. identification} - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract Systems and methods disclosed herein for providing people categorization within electronic images. One embodiment involves retrieving an input image from memory. The embodiment further involves determining one or more facial regions in the input image using a facial detection algorithm. The embodiment further involves identifying a number of features for each facial region. The embodiment further involves determining a subject face and a crowd face in the input image based on at least the number of features for the subject face being more than the number of features for the crowd face. The embodiment further involves displaying, on a display device, the subject face identified within the image. ## Claims ### Claim 1 A method that includes one or more processing devices performing operations comprising: retrieving a first input image from memory; determining two or more facial regions in the first input image using a facial detection algorithm; identifying features for each facial region of the two or more facial regions; determining, from the two or more facial regions in the first input image, a subject face and a crowd face in the first input image based on at least a number of the features for the subject face being more than a number of the features for the crowd face; causing a display device to display at least the subject face identified within the first input image; retrieving a second input image from the memory; determining two or more facial regions in the second input image using the facial detection algorithm; identifying features for each of the two or more facial regions in the second input image; determining, from the two or more facial regions in the second input image, a second subject face and a second crowd face in the second input image based on at least a number of the features for the second subject face being more than a number of the features for the second crowd face; and causing the display device to display at least the second subject face identified within the second input image. ### Claim 2 (depends on 1) The method of claim 1, wherein identifying the features for each facial region of the two or more facial regions in the first input image comprises resizing the facial region to a predetermined size. ### Claim 3 (depends on 2) The method of claim 2, wherein the features for each facial region of the two or more facial regions in the first input image are identified from a base layer of the facial region. ### Claim 4 (depends on 1) The method of claim 1, wherein determining the subject face and the crowd face further comprises: obtaining a predetermined threshold by performing k-mean clustering on the numbers of the features for the two or more facial regions in the first input image; and determining that the number of the features for the subject face is above the predetermined threshold, and the number of the features for the crowd face is below the predetermined threshold. ### Claim 5 (depends on 1) The method of claim 1, further comprising: determining that the crowd face has previously been identified as a previous subject face in a previous input image; and based on determining that the crowd face has previously been identified as the previous subject face, determining that the crowd face is a subject face in the first input image. ### Claim 6 (depends on 1) The method of claim 1, wherein identifying the features for each facial region of the two or more facial regions in the first input image is based on at least one of an Oriented Fast Rotated Brief feature detector, a Haar-like feature detector, or a Scale Invariant Feature Transform feature detector. ### Claim 7 A system comprising: a processor; and a non-transitory computer-readable medium communicatively coupled to the processor, wherein the processor is configured to execute program code stored in the non-transitory computer-readable medium and thereby perform operations comprising: retrieving a first input image from memory; determining two or more facial regions in the first input image using a facial detection algorithm; identifying features for each facial region of the two or more facial regions in the first input image; determining, from the two or more facial regions in the first input image, a subject face and a crowd face in the first input image based on at least a number of the features for the subject face being more than a number of the features for the crowd face; causing a display device to display at least the subject face identified within the first input image; retrieving a second input image from the memory; determining two or more facial regions in the second input image using a facial detection algorithm; identifying features for each facial region of the two or more facial regions in the second input image; determining, from the two or more facial regions in the second input image, a second subject face and a second crowd face in the second input image based on at least the number of the features for the second subject face being more than the number of the features for the second crowd face; and causing a display device to display at least the second subject face identified within the second input image. ### Claim 8 (depends on 7) The system of claim 7, wherein identifying the features for each facial region of the two or more facial regions in the first input image comprises resizing the facial region to a predetermined size. ### Claim 9 (depends on 8) The system of claim 8, wherein identifying the features for each facial region of the two or more facial regions in the first input image is based on a base layer of the facial region. ### Claim 10 (depends on 7) The system of claim 7, wherein determining the subject face and the crowd face further comprises obtaining a predetermined threshold based on k-mean clustering of the numbers of the features for the two or more facial regions in the first input image, and determining that the number of features for the subject face is above the predetermined threshold, and the number of features for the crowd face is below the predetermined threshold. ### Claim 11 (depends on 7) The system of claim 7, further comprising: determining that the crowd face has previously been identified as a previous subject face in a previous input image; and based on determining that the crowd face has previously been identified as the previous subject face, determining that the crowd face is a subject face in the first input image. ### Claim 12 (depends on 7) The system of claim 7, wherein identifying the features for each facial region of the two or more facial regions in the first input image is based on at least one of an Oriented Fast Rotated Brief feature detector, a Haar-like feature detector, or a Scale Invariant Feature Transform feature detector. ### Claim 13 A non-transitory computer-readable medium having program code that is stored thereon, the program code executable by one or more processing devices for performing operations comprising: accessing a first input image from a memory device; filtering the first input image to identify subject faces, wherein filtering the first input image comprises: determining two or more facial regions in the first input image using a facial detection algorithm; identifying features for each facial region of the two or more facial regions in the first input image; and determining, from the two or more facial regions in the first input image, a subject face and a crowd face in the first input image based on at least a number of the features for the subject face being more than a number of the features for the crowd face; causing a display device to display at least the subject face identified within the first input image; retrieving a second input image from the memory device; determining two or more facial regions in the second input image using the facial detection algorithm; identifying features for each facial region of the two or more facial regions in the second input image; determining a second subject face and a second crowd face in the second input image based on at least a number of the features for the second subject face being more than the number of the features for the second crowd face; and causing the display device to display at least the second subject face identified within the second input image. ### Claim 14 (depends on 13) The non-transitory computer-readable medium of claim 13, wherein identifying the features for each facial region of the two or more facial regions in the first input image comprises resizing the facial region to a predetermined size. ### Claim 15 (depends on 14) The non-transitory computer-readable medium of claim 14, wherein identifying the features for each facial region of the two or more facial regions in the first input image is based on a base layer of the facial region. ### Claim 16 (depends on 13) The non-transitory computer-readable medium of claim 13, wherein determining the subject face and the crowd face further comprises obtaining a predetermined threshold based on k-mean clustering of the numbers of the features for the two or more facial regions in the first input image, and determining that the number of the features for the subject face is above the predetermined threshold, and the number of the features for the crowd face is below the predetermined threshold. ### Claim 17 (depends on 13) The non-transitory computer-readable medium of claim 13, the program code further comprising: determining that the crowd face has previously been identified as a previous subject face in a previous input image; and based on determining that the crowd face has previously been identified as the previous subject face, determining that the crowd face is a subject face in the first input image.