# US 12,002,259 B2 · Image processing apparatus, training apparatus, image processing method, training method, and storage medium > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,002,259 B2 - Document ID: US12002259B2 - Status: Granted - Kind: B2 - Application number: 17/149,772 - Filed: 2021-01-15 - Published: 2021-05-06 - Granted: 2024-06-04 - Earliest claimed: 2017-07-05 - Assignees: Canon Kabushiki Kaisha - Applicants: Canon Kabushiki Kaisha - Inventors: Masato Aoba; Koichi Magai - Web page: https://patentgrep.com/patent/US12002259B2 ## CPC - G06V 10/82: using neural networks - G06F 18/2115: by evaluating different subsets according to an optimisation criterion, e.g. class separability, forward selection or backward elimination - G06F 18/2411: based on the proximity to a decision surface, e.g. support vector machines - G06F 18/2414: {Smoothing the distance, e.g. radial basis function networks [RBFN]} - G06F 18/2431: Multiple classes - G06T 7/11: Region-based segmentation - G06T 7/12: Edge-based segmentation - G06V 10/764: using classification, e.g. of video objects - G06V 30/19173: {Classification techniques} - G06V 30/194: References adjustable by an adaptive method, e.g. learning - G06T 2207/10024: Color image - G06T 2207/20021: Dividing image into blocks, subimages or windows - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30201: Face - G06V 2201/07: Target detection - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract There is provided with an image processing apparatus, for example, for image recognition. An extraction unit extracts a feature amount from a target image. An estimation unit estimates distribution of regions having attributes different from each other in the target image based on the feature amount. ## Claims ### Claim 1 An image processing apparatus comprising: one or more processors configurable to function as a plurality of units comprising: (1) an obtaining unit configured to obtain an input image; (2) an extraction unit configured to extract a feature from the input image; and (3) an estimation unit configured to estimate information corresponding to a number of pixels belonging to a particular class in the input image in a case where the feature extracted by the extraction unit is input, wherein the estimation unit estimates the information using parameters trained by using training images. ### Claim 2 (depends on 1) The apparatus according to claim 1, wherein the one or more processors further function as a control unit configured to control image capturing by an image capturing unit based on an estimation result by the estimation unit. ### Claim 3 (depends on 2) The apparatus according to claim 2, wherein the control unit is further configured to control the image capturing by the image capturing unit in terms of at least one of an exposure, focus, or white balance, based on the estimation result by the estimation unit. ### Claim 4 (depends on 1) The apparatus according to claim 1, wherein the estimation unit is further configured to estimate a ratio of (1) the number of pixels belonging to the particular class to (2) the number of all pixels of the input image. ### Claim 5 (depends on 1) The apparatus according to claim 1, wherein the estimation unit is further configured to estimate a proportion of an area belonging to at least one of a class of sky, a plant, or a skin, in the input image. ### Claim 6 (depends on 1) The apparatus according to claim 1, wherein the input image is a divided image obtained by dividing a whole image. ### Claim 7 (depends on 1) The apparatus according to claim 1, wherein the input image has a plurality of regions belonging to respective classes. ### Claim 8 (depends on 1) The apparatus according to claim 1, wherein the estimation unit uses parameters that have been trained by using (a) features extracted from the training images and (b) supervisory information representing information corresponding to area sizes belonging to the particular class in the training images, and wherein the supervisory information represents numeric values corresponding to total area sizes of the plurality of areas belonging to the particular class in the training images. ### Claim 9 (depends on 1) The apparatus according to claim 1, wherein the estimation unit is trained to output a numeric value corresponding to a total number of pixels of a plurality of areas belonging to the particular class in the input image as an output in relation to the input image that has the plurality of areas belonging to the particular class. ### Claim 10 (depends on 1) The apparatus according to claim 1, wherein the information corresponding to the number of pixels belonging to the particular class in the input image is used to calculate a ratio of an area size belonging to the particular class in the input image to a size of the input image. ### Claim 11 An image processing method comprising: obtaining an input image; extracting a feature from the input image; and estimating, using parameters trained by using training images, information corresponding to a number of pixels belonging to a particular class in the input image in a case where the feature extracted from the input image is input. ### Claim 12 A non-transitory computer-readable medium storing a program which causes a computer having a processor and a memory to perform a method comprising: obtaining an input image; extracting a feature from the input image; and estimating, using parameters trained by using training images, information corresponding to a number of pixels belonging to a particular class in the input image in a case where the feature extracted from the input image is input. ### Claim 13 A training apparatus comprising one or more processors, wherein the one or more processors are configurable to function as a plurality of units comprising: (1) a first obtaining unit configured to obtain a training image for training of an estimator; (2) an extraction unit configured to extract a feature from the training image; (3) a second obtaining unit configured to obtain, as supervisory information, information corresponding to a number of pixels belonging to a particular class in the training image; and (4) a training unit configured to perform the training of the estimator using a combination of the feature and the supervisory information. ### Claim 14 A training method comprising: obtaining a training image for training of an estimator; extracting a feature from the training image; obtaining, as supervisory information, information corresponding to an area size belonging to a particular class in the training image; and performing the training of the estimator using a combination of the feature and the supervisory information. ### Claim 15 (depends on 14) The method according to claim 14, wherein the information corresponding to an area size belonging to a particular class in the training image is information corresponding to a number of pixels belonging to a particular class in the training image. ### Claim 16 An image processing apparatus comprising: one or more processors configurable to function as a plurality of units comprising: (1) an obtaining unit configured to obtain a divided image obtained by dividing an input image; (2) an extraction unit configured to extract a feature from the divided image; and (3) an estimation unit configured to estimate information corresponding to a number of pixels belonging to a particular class in the divided image in a case where the feature extracted by the extraction unit is input, wherein the estimation unit estimates the information using parameters trained by using training images. ### Claim 17 (depends on 16) The apparatus according to claim 16, wherein the estimation unit estimates a ratio of the particular class to the divided image based on the information corresponding to the number of pixels belonging to the particular class in the divided image.