Abstract: Embodiments of the disclosure provide an image processing method and an image processing device. The method includes the following. A first image is obtained, and the first image is down-sampled into a second image. A first resolution parameter combination of each second pixel in the second image is obtained. A target pixel among the first pixels is selected, and second pixels corresponding to the target pixel among the second pixels are accordingly obtained. A comparison result is determined between the target pixel and each second pixel corresponding to the target pixel, and a candidate pixel among the second pixels corresponding to the target pixel is accordingly selected. A second resolution parameter combination of the target pixel is selected based on the first resolution parameter combination of the candidate pixel. The target pixel is converted into an output pixel based on the second resolution parameter combination.

Abstract: A method of processing an image by an electronic device, includes: obtaining a label image of a label including at least one symbol associated with handling of clothing item; obtaining, from the label image, text information associated with the handling of the clothing item; identifying, from the label image, the at least one symbol associated with the handling of the clothing item, based on the text information and the label image; and obtaining handling information related to the clothing item based on the identified at least one symbol.

Abstract: A measurement system includes a first feature point data generator, a second feature point data generator, and a calculator. The first feature point data generator generates first feature point data from first image data or from first shape data. The first image data is obtained from imaging of a measurement target and includes a predetermined portion of the measurement target. The first shape data is generated based on the first image data. The second feature point data generator generates second feature point data from second image data different from the first image data or from second shape data. The calculator calculates a positional correspondence of the predetermined portion of the measurement target between the first image data and the second image data or between the first shape data and the second shape data based on the first feature point data and the second feature point data.

Abstract: A method of improving a main output of a main AI model includes processing first video data using a machine learning model includes continuously analyzing incoming video data via a first processing pipeline and concurrently via a second processing pipeline. The first processing pipeline can include preprocessing incoming video data according to first preprocessing parameters that format the incoming video data to create the first video data and processing the first video data by the main AI model to determine the main output that is indicative of a first inference dependent upon the first video data. The second processing pipeline can include identifying first test preprocessing parameters that are predictive to produce a first test output that satisfies a baseline criterion and preprocessing the incoming video data according to the first test preprocessing parameters that format the incoming video data to create first test video data.

Abstract: Hyperspectral imaging (HSI) of biofluids for biometric applications is disclosed. A HSI device scans a region of a human subject's skin with sweat thereon, without requiring the human subject to touch any surface. The resulting HSI data is processed to identify aspects of the data that represent physical and/or chemical features that uniquely identify the subject. Processing the HSI data includes identifying one or more particular metabolites in the subject's sweat by analyzing wavelengths which those metabolites reflect. Processing the HSI data includes identifying one or more glands beneath the outermost surface of the subject's skin based on HSI-detected wavelength reflectance and wavelength penetration. The disclosed system may generate and store an identity profile corresponding to a human's unique identity features, as determined based on the HSI data processing. A subject's identity is verified by comparing subsequent HSI scans of his skin with sweat thereon to his stored identity profile.

Abstract: Systems and methods for face annotation are described. One or more of the systems and methods include receiving a plurality of annotated images, wherein each annotated image of the annotated images comprises a caption; cropping the annotated image based on a face detection algorithm to obtain a face crop; comparing the face crop to the caption corresponding to the annotated image to obtain a caption similarity score; and filtering the plurality of annotated images based on the caption similarity score to obtain a plurality of annotated face images.

Abstract: To utilize a result of image analysis, an image analysis apparatus 100 includes an analysis result acquisition unit 110 and a counting unit 111. The analysis result acquisition unit 110 acquires analysis information being information acquired by analyzing a still image and including an appearance attribute to which a person included in the still image belongs. The counting unit 111 counts, for each appearance attribute, the number of persons belonging to the appearance attribute in the still image by using the analysis information, and generates counting information indicating a result of the counting in units of still images.

Abstract: Methods, systems, and apparatus for receiving, from a provider server and at an image processing server, medical data including at least one image; determining, by the image processing server using a neural network, the at least one image within the medical data contains PII; in response to determining that the at least one image in the medical data contains PII, classifying, by the neural network, at least one medical characteristic of the image; generating, by the image processing server using the medical characteristic classification, at least one metadata tag that describes the medical characteristic; modifying, by the image processing server, the medical data to replace the image with the generated metadata tag; and transmitting, by the image processing server to a transaction server, the modified medical data.

Abstract: Systems and methods are described herein for processing electronic medical images to determine a first machine learning system, the first machine learning system having been trained to identify regions of electronic medical images; receive a plurality of electronic medical images, each of the electronic medical images being associated with one or more subcategories; determine a subset of the plurality of electronic medical images that are associated with only one subcategory of the one or more subcategories; provide the subset of the plurality of electronic medical images to the first machine learning system, the first machine learning system identifying regions within the subset of the plurality of electronic medical images associated with the subcategory; and train a second machine learning system, using the identified regions and the subset of the plurality of electronic medical images.

Abstract: The relationship between coordinates associated with the distribution of radiation rays emitted from the inside of a physical object and coordinates associated with the outer shape of this physical object is easily derived. Based on first 3D data (Da) in which the distribution of radiation rays emitted from the inside of a physical object (subject P) is represented in first coordinates (Ca), second 3D data (Db) in which the inside and the outer shape of the physical object are represented in second coordinates (Cb), and third 3D data (Dc) in which at least a part of the outer shape of the physical object is represented in third coordinates (Cc), a relationship (R) between the first coordinates and the third coordinates is derived.

Abstract: A system identifies and selects the best image to use to retrain a machine learning algorithm, thereby creating the best model. The best image is identified by determining the image, in an object identification, that has the most uncertainty. In addition to determining uncertainties for images, the system uses priority scores, groupings and orderings of subsets of images by the priority scores, computing a complement of recall or difficulty measure, and selecting images based on the difficulty measure.

Abstract: An image processing apparatus presents a result of performing defect detection on an image captured of an inspection target as case information. The apparatus comprises an acquisition unit that acquires the case information according to first information including an image on which the defect detection is to be performed and a presentation unit that presents the case information acquired by the acquisition unit.

Abstract: An image processing method and an image processing apparatus are provided. The image processing method includes: receiving a camera pose of a camera corresponding to a target scene; generating a piece of prediction information including either a color of an object included in the target scene or a density of the object, wherein the prediction information is generated by applying, to a neural network model, three-dimensional (3D) points on a camera ray formed based on the camera pose; sampling, among the 3D points, target points corresponding to a static object, wherein the sampling is based on the piece of prediction information; and outputting a rendered image corresponding to the target scene by projecting a pixel value corresponding to the target points onto the target scene and rendering the target scene onto which the pixel value may be projected.

Abstract: A machine learning device includes one or more processors, and one or more memories coupled to the one or more processors. The one or more processors are configured to cooperate with a program in the one or more memories to execute a process including acquiring image combinations multiple times, each of the image combinations including brightness images captured at different imaging positions and a training image associated with one of the brightness images, performing a manipulation process that includes manipulating the brightness images, and generating a machine learning model configured to receive the manipulated brightness images and the training image as input. The manipulation process includes a noise addition process that includes adding partial noise to different positions in the brightness images in an image combination of the image combinations.

Abstract: An image processing method according to the present disclosure includes the steps of: receiving at least one image data; assigning a plurality of feature points to the at least one image data; fitting the at least one image data to have at least one of a predetermined size and a predetermined angle based on at least one feature line generated by connecting at least two of the feature points; and designing a smile line by applying at least one template to a feature region generated by the feature points of the fitted image data.

Abstract: Systems and methods for generating optical character recognition (OCR) models configured to identify characters from a variety of different documents. The OCR models are based on a base model. One or more outside models can be tested to determine their effectiveness in supplementing the base model. When an outside model is effective it is incorporated into the base model. Generations of base models can be created that provide for additional functionality that is not present in the preceding model. A family of the generations of base models is maintained.

Abstract: A method of identifying synthetic media can include identifying a facial image in video or images, extracting a first set of features from the facial image, extracting a second set of features from the facial image, wherein the first set of features are different than the second set of features, inputting the first set of features into a first prediction model, generating a first output indicative of a nature of the facial image, inputting the second set of features into a second prediction model, generating a second output indicative of the nature of the facial image, and determining the nature of the facial image using the first output and the second output.

Abstract: A system and method for monitoring a switchgear includes an infrared camera, a processing unit, and an output unit. The infrared camera acquires an infrared image of the switchgear, and the processing unit converts it into a binary image. Pixels in the infrared image having a temperature equal to or above a first threshold value are given the same first value. Pixels in the infrared image having a temperature below the first threshold value are given the same second value. The processing unit is configured to implement a Siamese neural network to determine if a hot spot exists in the infrared image.

Abstract: A device for classifying image data includes a trainable pre-processing unit configured to retrieve, from a trained context, and based on the image data, at least one specification in terms of how a degree of granulation of the image data is to be reduced, and to reduce the degree of granulation of the image data in accordance with the at least one specification. The device further includes a trainable classifier configured to map the granulation-reduced image data onto an assignment to one or more classes of a specified classification.

Abstract: A method, device, system, and medium for classifying green-blue-gray infrastructure. The method includes obtaining a multispectral photo corresponding to a target area, and obtaining an image set of a target and a color orthophoto, based on the multispectral photo; obtaining a sample file based on the color orthophoto, and obtaining a classification result of the green-blue-gray infrastructure corresponding to the target area based on the image set of the target and the sample file. The embodiments can obtain the image set of the target and the color orthophoto based on the multispectral photo corresponding to the target area and combine the sample file obtained from the color orthophoto with the image set of the target to obtain the classification result of the green-blue-gray infrastructure corresponding to the target area, thereby improving the accuracy and efficiency for classifying the green-blue-gray infrastructure.