A method for tracking an object performed by an object tracking system includes encoding locations of visible objects in an environment captured in a current frame of a sequence of frames. The method also includes generating a representation of a current state of the environment based on an aggregation of the encoded locations and an encoded location of each object visible in one or more frames of the sequence of frames occurring prior to the current frame. The method further includes predicting a location of an object occluded in the current frame based on a comparison of object centers decoded from the representation of the current state to object centers saved from each prior representation associated with a different respective frame of the sequence of frames occurring prior to the current frame. The method still further includes adjusting a behavior of an autonomous agent in response to identifying the location of the occluded object.

An electronic device and a method of controlling an electronic device are provided. The method includes obtaining an image comprising a text through a camera, identifying an input text, among texts included in the image, to be translated, obtaining a first vector corresponding to the input text by inputting the identified input text to an encoder of a translation model, identifying whether additional information is necessary to translate the input text by inputting the first vector to a first artificial intelligence model trained to translate the input text, based on identification that the additional information is necessary, identifying additional information among the at least one context information by inputting the first vector and at least one context information obtained from the image, and obtaining an output text corresponding to the input text by inputting the first vector and the identified additional information to a decoder of the translation model.

The inventive concepts provide a method of providing a stochastic prediction system. The method includes extracting contours of patterns corresponding to a first design layout from a plurality of scanning electron microscope (SEM) images, respectively, generating a first contour histogram image based on the contours, and training a stochastic prediction model by using the first contour histogram image as an output, and by using the first design layout and a first resist image, a first aerial image, a first slope map, a first density map, and/or a first photo map corresponding to the first design layout as inputs, in which the stochastic prediction model comprises a cycle generative adversarial network (GAN).

An automated segmentation and identification system/method for identifying geographic atrophy (GA) phenotypic patterns in fundus autofluorescence images. A hybrid process combines a supervised pixel classifier with an active contour algorithm. A trained, machine learning model (e.g., SVM or U-Net) provides initial GA segmentation/classification, and this is followed by Chan-Vese active contour algorithm. The junctional zones of the GA segmented area are then analyzed for geometric regularity and light intensity regularity. A determination of GA phenotype is made, at least in part, from these parameters.

A system identifies anomalies in an image of an object. An input image of the object containing zero or more anomalies is supplied to an image encoder. The image encoder generates an image model. The image model is applied to an image decoder that forms a substitute non-anomalous image of the object. Differences between the input image and the substitute non-anomalous image identify zero or more areas of the input image that contain the zero or more the anomalies. The system implements a flow-based model and has been trained using (a) a set of augmented anomaly-free images of the object applied at the image encoder and (b) a reconstruction loss calculated based on a norm of differences between each augmented anomaly-free image of the object and a corresponding output image from the image decoder.

A first device may receive, from a second device, a request to approve a transaction wherein the request includes transaction data related to the transaction and an image of a signature of an individual that submitted the request. The first device may determine, after receiving the request, a priority level associated with the transaction based on the transaction data. The first device may process the image of the signature using a computer vision technique and/or a vector-based technique. The first device may select, from a memory storing a plurality of comparator signatures, a comparator signature for the signature based on the priority level. The first device may use the comparator signature to verify the signature to approve or deny the transaction. The first device may perform a comparison of the comparator signature and the signature in the image after processing the image and selecting the comparator signature.

In implementations of vector-based glyph style transfer, a style transfer system transfers a modification of a modified glyph to an additional glyph. The system identifies the modification by comparing the modified glyph to a corresponding unmodified glyph. In one or more implementations, this includes identifying similar segments of the additional glyph based on a style transfer policy, which defines conditions for transferring the modification to the additional glyph. The system transfers the modification to the additional glyph by mapping the modification to the similar segments.

A system and computer-implemented method are provided for preprocessing medical image data for machine learning. Image data is accessed which comprises an anatomical structure. The anatomical structure in the image data is segmented to obtain a segmentation of the anatomical structure as a delineated part of the image data. A grid is assigned to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, wherein said assigning comprises adapting the grid to fit the segmentation of the anatomical structure in the image data. A machine learning algorithm is then provided with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid. In some embodiments, the image data of the anatomical structure may be resampled using the assigned grid. Advantageous, a standardized addressing to the image data of the anatomical structure is provided, which may reduce the computational overhead of the machine learning, require fewer training data, etc.

A system and computer-implemented method are provided for preprocessing medical image data for machine learning. Image data is accessed which comprises an anatomical structure. The anatomical structure in the image data is segmented to obtain a segmentation of the anatomical structure as a delineated part of the image data. A grid is assigned to the delineated part of the image data, the grid representing a partitioning of an exterior and interior of the type of anatomical structure using grid lines, wherein said assigning comprises adapting the grid to fit the segmentation of the anatomical structure in the image data. A machine learning algorithm is then provided with an addressing to the image data in the delineated part on the basis of coordinates in the assigned grid. In some embodiments, the image data of the anatomical structure may be resampled using the assigned grid. Advantageous, a standardized addressing to the image data of the anatomical structure is provided, which may reduce the computational overhead of the machine learning, require fewer training data, etc.

A system for vector extraction comprising a vector extraction engine stored and operating on a network-connected computing device that loads raster images from a database stored and operating on a network-connected computing device, identifies features in the raster images, and computes a vector based on the features, and methods for feature and vector extraction.