Abstract: A method is disclosed. The method may include receiving an image or video; extracting a plurality of features from the image or video; executing a neural network using the plurality of features to obtain a performance score for the image or video, the neural network comprising an input layer, a plurality of intermediate layers subsequent to the input layer, and a regression layer or a classification layer; extracting values from one or more signals between an intermediate layer and the regression layer or the classification layer; for each of the plurality of features, calculating, based on at least one of the one or more values, an impact score indicating an impact the feature had on the performance score; and generating, based on one or more impact scores for the plurality of features, indications indicating an impact different features of the image or video had on the performance score.

Abstract: A method for context based separation between vehicle pixels and background pixels, which may include receiving a first image and a second image, the first image and the second images are of temporarily adjacent to each other and capture a same vehicle and background content. The method may also include obtaining a first bounding box that surrounds at least a part of the vehicle within the first image; wherein the first bounding has a first width and a first height; obtaining a second bounding box that surrounds at least the part of the vehicle within the second image; wherein the second bounding box has a second width and a second height; obtaining a mapping between pairs of initially matched pixels, wherein each pair comprises a first pixel and a second pixel that correspond to same entity portion, wherein the entity portion is a portion of the vehicle or of the background content.

Abstract: Methods, systems, and apparatuses for image segmentation are provided. For example, a computing device may obtain an image, and may apply a process to the image to generate input image feature data and input image segmentation data. Further, the computing device may obtain reference image feature data and reference image classification data for a plurality of reference images. The computing device may generate reference image segmentation data based on the reference image feature data, the reference image classification data, and the input image feature data. The computing device may further blend the input image segmentation data and the reference image segmentation data to generate blended image segmentation data. The computing device may store the blended image segmentation data within a data repository. In some examples, the computing device provides the blended image segmentation data for display.

Abstract: A method of characterizing a specimen to be analyzed in an automated diagnostic analysis system provides an HILN classification (hemolysis, icterus, lipemia, normal) of the specimen along with a basis for that determination. The method includes assigning a hash code to each training image of a sample specimen used in the characterization training process. In response to an HILN determination for a test specimen, the method can retrieve via the hash code one or more of the closest matching training images upon which the HILN classification is based. The one or more of the closest matching training images can be displayed alongside of the one or more images of the test specimen. Quality check modules and systems configured to carry out the method are also described, as are other aspects.

Abstract: A device and a corresponding system are disclosed. The device may comprise a display. Additionally, the device and/or system may be operable to detect stale images. Further, the stale images may be detected in devices and/or systems where incoming images—such as from an imager—are received at a frame rate different than that of the display. A controller may detect the stale images by assigning hash values to images of each frame displayed by the display; storing novel hash values in a memory; and maintaining a counter. The counter may be maintained such that assigning a repeat hash value increments the counter by one and assigning novel hash values resets the counter to zero. Further, the stale images may be determined based, at least in part, on the counter value.

Abstract: The invention relates to a computer-implemented method for automatically detecting anatomical structures (3) in a medical image (1) of a subject, the method comprising applying an object detector function (4) to the medical image, wherein the object detector function performs the steps of: (A) applying a first neural network (40) to the medical image, wherein the first neural network is trained to detect a first plurality of classes of larger-sized anatomical structures (3a), thereby generating as output the coordinates of at least one first bounding box (51) and the confidence score of it containing a larger-sized anatomical structure; (B) cropping (42) the medical image to the first bounding box, thereby generating a cropped image (11) containing the image content within the first bounding box (51); and (C) applying a second neural network (44) to the cropped medical image, wherein the second neural network is trained to detect at least one second class of smaller-sized anatomical structures (3b), thereby g

Abstract: Methods and systems are disclosed for generating training data for a machine learning model for better performance of the model. A source image is selected from an image database, along with a target image. An image segmenter is utilized with the source image to generate a source image segmentation mask having a foreground region and a background region. The same is performed with the target image to generate a target image segmentation mask having a foreground region and a background region. Foregrounds and backgrounds of the source image and target image are determined based on the masks. The target image foreground is removed from the target image, and the source image foreground is inserted into the target image to create an augmented image having the source image foreground and the target image background. The training data for the machine learning model is updated to include this augmented image.

Abstract: A multi-task text infilling system receives a digital image and identifies a region of interest of the image that contains original text. The system uses a machine learning model to determine, in parallel: a foreground image that includes the original text; a background image that omits the original text; and a binary mask that distinguishes foreground pixels from background pixels, The system receives a target mask that contains replacement text. The system then applies the target mask to blend the background image with the foreground layer image and yield a modified digital image that includes the replacement text and omits the original text.

Abstract: In one aspect, an example method includes a processor (1) applying a feature map network to an image to create a feature map comprising a grid of vectors characterizing at least one feature in the image and (2) applying a probability map network to the feature map to create a probability map assigning a probability to the at least one feature in the image, where the assigned probability corresponds to a likelihood that the at least one feature is an overlay. The method further includes the processor determining that the probability exceeds a threshold, and responsive to the processor determining that the probability exceeds the threshold, performing a processing action associated with the at least one feature.

Abstract: According an embodiment, a computing device can: identify, in a chat view associated with a video chat session, a first authorized participant and a second authorized participant of the video chat session; render, in the chat view, first visual data indicative of the first authorized participant and second visual data indicative of the second authorized participant based at least in part on identification of the first authorized participant and the second authorized participant, respectively; define, in the chat view, a chat zone indicative of a reference location of the first authorized participant; determine that the first authorized participant moved outside the chat zone; and/or conceal, in the chat view, the first visual data indicative of the first authorized participant based at least in part on determination that the first authorized participant moved outside the chat zone.

Abstract: Techniques for mapping size information associated with a client to target brands, garments, sizes, shapes, and styles for which there is no standardized correlation. The size information associated with a client may be generated by modeling client garments, accessing computer aided drawing (CAD) files associated with client garments, or by analyzing a history of garment purchases associated with the client. Information for target garments may be generated in a similar fashion. A system may then create a standardized scale with a set of sizes for a target, and map a client base size to that standardized size scale. Similar matching and mapping may also be done with shape and style considerations. A recommendation based on the mapping may then be communicated to the client.

Abstract: An image processing method and apparatus, a device, and a storage medium. The image processing method includes: performing preliminary segmentation recognition on a raw image by using a first segmentation model to obtain a candidate foreground image region and a candidate background image region of the raw image; recombining the candidate foreground image region, the candidate background image region, and the raw image to obtain a recombined image, pixels in the recombined image being in a one-to-one correspondence with pixels in the raw image; and performing region segmentation recognition on the recombined image by using a second segmentation model to obtain a target foreground image region and a target background image region of the raw image.

Abstract: A video summary device may determine, based on video data, a plurality of events that occurred during a period of time. The video summary device may determine first measures of relevance of a plurality of portions of the period of time. A first measure of relevance of a portion of time may be determined based on one or more events of the plurality of events. The video summary device may determine second measures of relevance of a plurality of ranges of time between different portions of time of the plurality of portions of the period of time. The second measures of relevance is determined based on the first measures of relevance determined for the different portions of time. The video summary device may determine particular frames of the video data, as a video summary of the video data, based on the second measures of relevance.

Abstract: An image processing system generates an image mask from an image. The image is processed by an object detector to identify a region having an object, and the region is classified based on an object type of the object. A masking pipeline is selected from a number of masking pipelines based on the classification of the region. The region is processed using the masking pipeline to generate a region mask. An image mask for the image is generated using the region mask.

Abstract: This application relates to the field of digital image processing technologies, and discloses an image-text fusion method and apparatus, and an electronic device, to minimize blockage of a saliency feature in an image by a text when the text is laid out in the image, and obtain a higher visual balance degree after the text is laid out in the first image, thereby achieving a better layout effect. According to the method of this application, first, a plurality of candidate text templates and layout positions of a plurality of corresponding texts in an image can be determined, so that a text laid out in the image does not block a visually salient object having a greater feature value, such as a human face or a building.

Abstract: The present invention relates to the field of three-dimensional images, and provides a method for three-dimensional reconstruction of a target object, comprising the following steps: S10: calculating and acquiring a visual angle boundary of the target object; S20: acquiring image information of the target object under multiple visual angles in the visual angle boundary, the image information including initial image information and/or fused image information; S30: finishing three-dimensional reconstruction of the target object according to the acquired image information. By adopting the method for three-dimensional reconstruction of the target object in the present invention, three-dimensional reconstruction of the three-dimensional target at different moments can be carried out in real time, no other links are carried out during the operation, the acquired image information is not lost, real information is directly reproduced, and three-dimensional reconstruction of the real target object is realized.

Abstract: An example of an apparatus to infer joint rotations and positions is provided. The apparatus includes a communications interface to receive raw data from an external source. The raw data includes a first joint position and a second joint position of an input skeleton. In addition, the apparatus includes a memory storage unit to store the raw data. Furthermore, the apparatus includes a pre-processing engine to generate normalized data from the raw data. The normalized data is to be stored in the memory storage unit. Also, the apparatus includes an inverse kinematics engine to apply a neural network to infer a joint rotation from the normalized data. The neural network is to use historical data. Training data used to train the neural network includes positional noise.

Abstract: Systems and methods for multi-task joint training of a neural network including an encoder module and a multi-headed attention mechanism are provided. In one aspect, the system includes a processor configured to receive input data including a first set of labels and a second set of labels. Using the encoder module, features are extracted from the input data. Using a multi-headed attention mechanism, training loss metrics are computed. A first training loss metric is computed using the extracted features and the first set of labels, and a second training loss metric is computed using the extracted features and the second set of labels. A first mask is applied to filter the first training loss metric, and a second mask is applied to filter the second training loss metric. A final training loss metric is computed based on the filtered first and second training loss metrics.

Abstract: Methods and systems are described for estimating an optimal pupil binarization threshold using adaptive binarization thresholding. The disclosed methods and systems are designed to determine a pupil binarization threshold from a histogram of an eye image. An eye image is obtained and an eye image histogram is computed from the eye image. A pupil region and an iris region are identified in the eye image histogram and the second derivative of the eye image histogram is computed. The pupil binarization threshold is determined based on the second derivative of the eye image histogram, the identified pupil region and the identified iris region and then used to generate a binarized eye image. A pupil contour may be determined from the binarized eye image. The disclosed system may help to overcome challenges associated with robust and precise detection of a pupil contour, for example, in images of varying quality.

Abstract: A method and system determine network based access to restricted systems. The method includes receiving a request for a permission access status of a party seeking access to one of the restricted systems. A database of periodically updated lists of entities is accessed. A name of the party is extracted from the request. A determination is made whether the name does not match one of the entities. The name is decomposed into parts if the name not matching one of the entities. A determination is made whether any of the parts of the name matches one of the entities. A denial of access status is forwarded from the computer server to an external computing device if any of the parts of the name matches one of the entities.

Abstract: A system and method are capable of ensuring that one or more text strings will be able to be fully rendered in a target area of a user interface or a target area of a graphics file. The system and method determine the number of pixels of first and second reference text that fit in the target area in the horizontal direction and the vertical direction, respectively, determine the number of pixels of string text in the horizontal direction and the vertical direction, and compare the number of pixels in the horizontal direction of the first reference text and the vertical direction of the second reference text respectively to the number of pixels in the horizontal direction and the vertical direction of the text string that is desired to be rendered in the target area to determine whether the text string will fit in the target area.

Abstract: This disclosure describes one or more implementations of an alpha matting system that utilizes a deep learning model to generate alpha mattes for digital images utilizing an alpha-range classifier function. More specifically, in various implementations, the alpha matting system builds and utilizes an object mask neural network having a decoder that includes an alpha-range classifier to determine classification probabilities for pixels of a digital image with respect to multiple alpha-range classifications. In addition, the alpha matting system can utilize a refinement model to generate the alpha matte from the pixel classification probabilities with respect to the multiple alpha-range classifications.

Abstract: In an image processing method, a detection image and a marked image are obtained. An image segmentation model is applied to segment a first segmented image from the detection image. The first segmented image is corrected according to the marked image to obtain a second segmented image. A size of the second segmented image is adjusted to obtain an adjusted segmented image. The adjusted segmented image is used as a standard segmented image of the detection image. The method improves accuracy of image segmentation and recognition.

Abstract: Methods, systems, and apparatus, including medium-encoded computer program products, for computer aided design of physical structures using generative design processes include: obtaining one or more design criteria, a model of an object for which a physical structure is to be manufactured, and a set of eigenmodes from a modal analysis of the model of the object; extracting a proper subset of non-zero eigenmodes from the set of eigenmodes, wherein the proper subset of non-zero eigenmodes include at least three lowest valued, non-zero eigenmodes; combining data of the proper subset of non-zero eigenmodes to form a strain energy field for the model of the object; iteratively modifying a generatively designed shape of the model of the object using the strain energy field to drive changes to the generatively designed shape of the model; and providing the generatively designed shape of the model of the object.

Abstract: A system identifying anomalies in an image of an object is first trained using first sets of images corresponding to first anomaly types for the object. A model of the object is formed in a latent space. A label for each anomalous image is used to calculate vectors containing means and standard deviations for each first anomaly types. The means and standard deviations are used to calculate a log-likelihood loss for each first anomaly type. The system is retrained using second sets of images corresponding to second anomaly types for the object. The vectors are supplemented using labels for each second anomaly types. A statistically sufficient sample of information in the means and standard deviations vectors is supplied to the latent space. A log-likelihood loss for each of the first and second anomaly types is calculated based on their respective mean and standard deviation.

Abstract: A max-flow/min-cut solution algorithm for early terminating a push-relabel algorithm is provided. The max-flow/min-cut solution algorithm is used for an application that does not require an exact maximum flow, and includes: defining an early termination condition of the push-relabel algorithm by a separation condition and a stable condition; determining that the separation condition is satisfied if there is no source node s, s?S, in the set T at any time in an operation process of the push-relabel algorithm; determining that the stable condition is satisfied if there is no active node in the set T; and terminating the push-relabel algorithm if both the separation condition and the stability condition are satisfied. The early termination technique is proposed to greatly reduce redundant computations and ensure that the algorithm terminates correctly in all cases.

Abstract: A method for spatial resolution enhanced imaging is provided. The method includes determining a target region of a subject being imaged; estimating an achievable spatial resolution; selecting an increased spatial resolution that is greater than a predetermined spatial resolution and equal to or less than the achievable spatial resolution; and acquiring image data of the target region with the increased spatial resolution. A method for enhancing the contrast between the region of interest and other tissue region in the image data acquired with the increased spatial resolution and other tissue region in image data acquired with the predetermined spatial resolution due to the intrinsic partial volume effect is also provided.

Abstract: A method and apparatus are provided for detecting respective potential presences of respective different cellular fluorescence pattern types on a biological cellular substrate including human epithelioma cells (HEp cells), wherein the fluorescence pattern types include different antinuclear antibody fluorescence pattern types. A method is also provided for detecting potential presences of different cellular fluorescence pattern types on a biological cellular substrate including human epithelioma cells by means of digital image processing, as well as a computing unit, a data network device, a computer program product and a data carrier signal therefor.

Abstract: An image segmentation method includes the following steps: obtaining a target image; inputting the target image into a machine learning model to obtain an image segmentation parameter value corresponding to the target image; executing an image segmentation algorithm on the target image according to the image segmentation parameter value to obtain an image segmentation result, wherein the image segmentation result is segmenting the target image into object regions; and displaying the image segmentation result. In addition, an electronic device and storage medium using the method are also provided.

Abstract: A method, computer program, and computer system are provided for image segmentation. Image data, such as biological image data, is received. One or more objects associated with the received image data is detected. One or more regions of interest are determined within the receive image data corresponding to one or more segments based on the detected objects.

Abstract: A system, method, and/or computer storage medium encoded with a computer program are disclosed for providing an interactive virtual event session for respective pluralities of devices substantially in real-time. A data communication session is provided, to which a plurality of computing devices operated by participants of an event session connect. Respective audio-video feeds including content captured by respective cameras and microphones configured with the plurality of computing devices are received. Moreover, at least some of the content in at least two of the respective audio-video feeds is adjusted, including by executing editing processes. Furthermore, a virtual setting for the event session is provided to each of the plurality of computing devices, and modified to include at least the adjusted content. Changing views of the adjusted content and the virtual setting is provided to each of the plurality of computing devices, using artificial intelligence.

Abstract: A device including a field programmable gate array is configured to set a coherent clock time frame for data captured. The gate array creates metadata that is operable for capturing a state of each of a plurality of faces of an information space. A geodesic lens comprises a geodesic equation that creates metadata from the captured data. An anamorphic display engine is used to transduce a plurality of visual dimensions of related foreground and background objects and render the objects relative motion in time.

Abstract: A method includes receiving, from a camera, one or more frames of image data of a scene comprising a background and one or more three-dimensional objects, wherein each frame comprises a raster of pixels of image data; detecting layer information of the scene, wherein the layer information is associated with a depth-based distribution of the pixels in the one or more frames; and determining a multi-layer model for the scene, the multi-layer model comprising a plurality of discrete layers comprising first and second discrete layers, wherein each discrete layer is associated with a unique depth value relative to the camera. The method further includes mapping the pixels to the layers of the plurality of discrete layers; rendering the pixels as a first image of the scene as viewed from a first perspective; and rendering the pixels as a second image of the scene as viewed from a second perspective.

Abstract: A method for driving vibration based on micro-touch, including: acquiring, according to a given input of a touch interface, a plurality of node coordinates of a dynamic effect model corresponding to the given input and a total duration of the node coordinates; generating a dynamic effect curve according to the plurality of node coordinates and the total duration; obtaining a vibration characteristic curve from the dynamic effect curve according to a preset mapping rule; and generating a vibration driving file based on the vibration characteristic curve. The vibration driving file is configured to drive a vibrating motor to vibrate. An apparatus for driving vibration, an electronic device, and a computer-readable storage medium are also provided. The method for driving vibration based on micro-touch and the related device can achieve generation of an appropriate micro-touch interaction effect according to animation matching of application scenes, and bring better user experience.

Abstract: Object parts (20, 21, 22, 23, 24) are detected in a picture using object detector(s) (3) and part location representations (40, 42, 43, 44) are generated for the detected object parts (20, 22, 23, 24). The size of an object (10) comprising object parts (20, 21, 22, 23, 24) is estimated based on a geometric model and the part location representations (40, 42, 43, 44). Search locations (51) in the picture for a search window (52) having a size based on the estimated size are determined based on the part location representations (40, 42, 43, 44). The search locations (51) are then processed by identifying any detected object part (20, 22, 23) that is within the search window (52) positioned at the search location (51). A homography is estimated by minimizing an error between mapped object part(s) from the geometric model and the identified detected object part(s) (20, 22, 23). If the error is smaller than a threshold value, an object location representation is determined for the object (10).

Abstract: A computer-implemented method for segmenting an object in at least one image acquired by a camera including computing an edge probabilities image based on the image, said edge probabilities image comprising, for each pixel of the image, the probability that said pixel is an edge, computing a segmentation probabilities image based on the image (IM), said segmentation probabilities image comprising, for each pixel of the image (IM), the probability that said pixel belongs to the object (OBJ), and computing a binary mask of the object based on the edge probabilities image and based on the segmentation probabilities image.

Abstract: Disclosed are a high-precision semi-automatic image data labeling method, an electronic apparatus and a non-transitory computer-readable storage medium. The high-precision semi-automatic image data labeling method may include: displaying a to-be-labeled image, the to-be-labeled image comprising a selected area and an unselected area; acquiring a coordinate point of the unselected area and a first range value; executing a grabcut algorithm based on the coordinate point of the unselected area and the first range value acquired, and obtaining a binarized image divided by the grabcut algorithm; executing an edge tracking algorithm on the binarized image to acquire current edge coordinates; updating a local coordinate set based on the current edge coordinates acquired; updating the selected area of the to-be-labeled image based on the local coordinate set acquired.

Abstract: Apparatuses, systems, and techniques to identify a shape or camera pose of a three-dimensional object from a two-dimensional image of the object. In at least one embodiment, objects are identified in an image using one or more neural networks that have been trained on objects of a similar category and a three-dimensional mesh template.

Abstract: One or more information maps are obtained by an agricultural work machine. The one or more information maps map one or more agricultural characteristic values at different geographic locations of a field. An in-situ sensor on the agricultural work machine senses an agricultural characteristic as the agricultural work machine moves through the field. A predictive map generator generates a predictive map that predicts a predictive agricultural characteristic at different locations in the field based on a relationship between the values in the one or more information maps and the agricultural characteristic sensed by the in-situ sensor. The predictive map can be output and used in automated machine control.

Abstract: A core-line extraction unit extracts a core line of the blood vessel from a three-dimensional image of a subject including the blood vessel. A local-route extraction unit extracts a local route of the core line within a predetermined range based on a reference point on the core line. In a case where a plurality of points on the local route are projected, a plane setting unit sets a plane having an orientation in which a distribution of a plurality of the projected points has a reference value based on a maximum value of the distribution, as a display plane of the three-dimensional image.

Abstract: Examples are disclosed relating to performing signal processing on time-of-flight sensor data using pixelwise temporal metrics. One example provides a computing system comprising a logic machine, and a storage machine holding instructions executable by the logic machine to obtain temporal phase data for a plurality of pixels as acquired by a time-of-flight image sensor, the temporal phase data comprising phase data for a plurality of light modulation frequencies, determine temporal active brightness data for the pixel, and, for each pixel of the plurality of pixels, determine a statistical metric for the temporal active brightness data. The instructions are further executable to perform phase unwrapping on the temporal phase data for the plurality of pixels to obtain a depth image, based on the statistical metric for the temporal active brightness data, perform a denoising operation on at least some pixels of the depth image, and output the depth image.

Abstract: A computer-implemented method comprising: receiving a 3D image including an object depicted in the image, the 3D image comprising an ordered set of 2D images; determining a contour around the object in a first of said 2D images; and determining a contour around the object in a second of said 2D images, the second 2D image being non-contiguous with the first in said ordered set, having an intermediate region comprising one or more intermediate ones of said 2D images between the first and second 2D images within said ordered set. In each of the first and second 2D images, inside of the contour is classified as foreground and outside of the contour is classified as background. The method further comprises performing a 3D geodesic distance computation to classify points in the intermediate region as foreground of background.

Abstract: An image sensor according to one embodiment may comprise: an image sensing unit which receives light and generates a first Bayer pattern having a first resolution; and a processor which receives a second Bayer pattern that is at least a portion of the first Bayer pattern from the image sensing unit, and then generates a third Bayer pattern having a higher resolution than the second alignment unit on the basis of the received second Bayer pattern.

Abstract: Imagery-based boundary identification for agricultural fields is provided. In various embodiments, a time series of surface reflectance rasters for a geographic region is received. For each of the surface reflectance rasters, at least one index raster is determined, yielding at least one time series of index rasters. The at least one time series of index rasters is divided into a plurality of consecutive time windows. The at least one time series of index rasters is composited within each of the plurality of time windows, yielding a composite index raster for each of the at least one time series of index rasters in each of the plurality of time windows. The composite index rasters are segmented into a plurality of spatially compact regions of the geographic region. A plurality of polygons is generated from the plurality of spatially compact regions, each of the plurality of polygons corresponding to an agricultural field in the geographic region.

Abstract: Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing a plurality of neural networks in a multi-branch pipeline to generate image masks for digital images. Specifically, the disclosed system can classify a digital image as a portrait or a non-portrait image. Based on classifying a portrait image, the disclosed system can utilize separate neural networks to generate a first mask portion for a portion of the digital image including a defined boundary region and a second mask portion for a portion of the digital image including a blended boundary region. The disclosed system can generate the mask portion for the blended boundary region by utilizing a trimap generation neural network to automatically generate a trimap segmentation including the blended boundary region. The disclosed system can then merge the first mask portion and the second mask portion to generate an image mask for the digital image.

Abstract: A rendered image is generated from a semiconductor device design file. The rendered image is segmented based on a grey level of the rendered image. Care areas are determined based on the segmenting. Defect inspection is performed in the care areas. This process can be performed on a wafer inspection tool that uses photon optics or electron beam optics.

Abstract: There is disclosed a computer-implemented method for lossy image or video compression, transmission and decoding, the method including the steps of: (i) receiving an input image at a first computer system; (ii) encoding the input image using a first trained neural network, using the first computer system, to produce a latent representation; (iii) quantizing the latent representation using the first computer system to produce a quantized latent; (iv) entropy encoding the quantized latent into a bitstream, using the first computer system; (v) transmitting the bitstream to a second computer system; (vi) the second computer system entropy decoding the bitstream to produce the quantized latent; (vii) the second computer system using a second trained neural network to produce an output image from the quantized latent, wherein the output image is an approximation of the input image. Related computer-implemented methods, systems, computer-implemented training methods and computer program products are disclosed.

Abstract: A method for enriching a historical learning base. Training of a first neural network to transform first photos of first dental scenes into first hyper-realistic images to simulate the effect of a dental event on the first dental scenes. Submission, to the trained first neural network, of a source photo representing a source dental scene in a dental context, so as to obtain a final image representing, hyper-realistically, the source dental scene after simulation of the dental event. Creation of a descriptor of the final image, or “final descriptor.” Creation of a historical record consisting of the final image and the final descriptor, and addition of the historical record in the historical learning base.

Abstract: An information processing apparatus (2000) includes a recognizer (2020). An image (10) is input to the recognizer (2020). The recognizer (2020) outputs, for a crowd included in the input image (10), a label (30) describing a type of the crowd and structure information (40) describing a structure of the crowd. The structure information (40) indicates a location and a direction of an object included in the crowd. The information processing apparatus (2000) acquires training data (50) which includes a training image (52), a training label (54), and training structure information (56). The information processing apparatus (2000) performs training of the recognizer (2020) using the label (30) and the structure information (40), which are acquired by inputting the training image (52) with respect to the recognizer (2020, and the training label (54) and the training structure information (56).

Abstract: An information processing apparatus (2000) includes a recognizer (2020). An image (10) is input to the recognizer (2020). The recognizer (2020) outputs, for a crowd included in the input image (10), a label (30) describing a type of the crowd and structure information (40) describing a structure of the crowd. The structure information (40) indicates a location and a direction of an object included in the crowd. The information processing apparatus (2000) acquires training data (50) which includes a training image (52), a training label (54), and training structure information (56). The information processing apparatus (2000) performs training of the recognizer (2020) using the label (30) and the structure information (40), which are acquired by inputting the training image (52) with respect to the recognizer (2020, and the training label (54) and the training structure information (56).