Abstract: An example apparatus for semantic image segmentation includes a receiver to receive an image to be segmented. The apparatus also includes a gated dense pyramid network comprising a plurality of gated dense pyramid (GDP) blocks to be trained to generate semantic labels for each pixel in the received image. The apparatus further includes a generator to generate a segmented image based on the generated semantic labels.

Abstract: The present technology relates to an information processing device, an information processing method, and a program that are designed to enable easy generation of a path for successively displaying characteristic images. The information processing device includes a setting unit that sets a path for connecting characteristic portions in at least one image by referring to metadata including at least information about feature points detected from the image. The setting unit sets the path by determining a regression curve, using the feature points. In a case where the feature points include a feature point at a distance equal to or longer than a predetermined threshold value from the regression curve, the setting unit redetermines the regression curve after removing a feature point detected from an image including a feature point having a low score among the feature points. The present technology can be applied to information processing devices that process still images and moving images.

Abstract: A method for determining whether a goal is achieved by a trajectory of a ball using a mobile computer device comprises capturing a sequence of video frames of the ball with a camera of the mobile computer device; detecting the ball in at least three of the video frames; computing a trajectory of the ball using the detections of the ball; detecting a goal image in at least one of the video frames; computing whether the trajectory of the ball achieves intersection or similar with a goal plane computed from the goal image according to a goal criterion.

Abstract: This invention provides a system and method that employs reduction in bandwidth and the amount of data stored, along with queuing of data, so that the significant and/or relevant shipboard visual information is detected and communicated continuously from the ship (or other remote vehicle/location) to shore without loss of useful/high-priority information and within the available bandwidth of that typical, available satellite link. The system and method supports remote configuration and management from shore to the ship over the same communications channel but in the reverse direction.

Abstract: An information processing device is configured to: obtain sensing data from an optical sensor; obtain position information of a mobile body which includes the optical sensor; obtain map information; determine, in sensing data, a specific sensing data region corresponding to a specific region in a sensing region of the optical sensor by using the position information and the map information that have been obtained; determine input information to be provided to an object detection model, according to the specific sensing data region; and cause the object detection model to perform object detection processing by using the input information.

Abstract: A method and system are disclosed for monitoring passengers in within a cabin of a vehicle and determining whether the passengers are engaging in abnormal behavior. The method and system uses a novel vector to robustly and numerically represent the activity of the passengers in a respective frame, which is referred to herein as an “activity vector.” Additionally, a Gaussian Mixture Model is utilized by the method and system to distinguish between normal and abnormal passenger behavior. Cluster components of the Gaussian Mixture Model are advantageously learned using an unsupervised approach in which training data is not labeled or annotated to indicate normal and abnormal passenger behavior. In this way, the Gaussian Mixture Model can be trained at a very low cost.

Abstract: A method for detecting objects in traffic flow includes identifying a location of a first camera and a second camera. The method includes determining a traffic flow direction. The method includes receiving a first image set captured by the first camera and a second image set captured by the second camera. The method includes identifying an object of a first object type in the first image set based on the location of the first camera and the traffic flow direction. The method includes identifying an object of a second object type in the second image set based on the location of the second camera and the traffic flow direction. The method includes recording the object of the first object type as a single incremented count in a first object type queue and the object of the second object type as a single incremented count in a second object type queue.

Abstract: An image processing apparatus includes at least one memory configured to store instructions; and at least one processor in communication with the at least one memory and configured to execute the instructions to: acquire a plurality of images having angles of view that at least partially overlap each other and varying in in-focus positions, and information corresponding to each of the plurality of images; determine a rearrangement method for rearranging the plurality of images in order of the in-focus positions based on the information; rearrange the plurality of images using the rearrangement method; and generate a combined image by extracting areas in focus from the plurality of images and performing the combining based on the extracted areas.

Abstract: A computing device, including a processor configured to receive a first video including a plurality of frames. For each frame, the processor may determine that a target region of the frame includes a target object. The processor may determine a surrounding region within which the target region is located. The surrounding region may be smaller than the frame. The processor may identify one or more features located in the surrounding region. From the one or more features, the processor may generate one or more manipulated object identifiers. For each of a plurality of pairs of frames, the processor may determine a respective manipulated object movement between a first manipulated object identifier of the first frame and a second manipulated object identifier of the second frame. The processor may classify at least one action performed in the first video based on the plurality of manipulated object movements.

Abstract: A method of analyzing film on a substrate comprises receiving surface profile data obtained from measurements of a plurality of discrete regions on a substrate, the plurality of discrete regions comprising one or more film layers; extracting a plurality of parameters from the received surface profile data, the plurality of parameters comprising one or more parameters of the one or more film layers of each of the plurality of discrete regions, wherein the extracting is based on a predetermined pattern for the plurality of the discrete regions on the substrate; and displaying a user interface.

Abstract: A computer implemented method includes obtaining a first deep neural network (DNN) model trained on labeled real image data for a downstream vision task, obtaining a second DNN model trained on synthetic images created with random image parameter values for the downstream vision task, obtaining a third DNN model trained on the labeled real image data and the synthetic images for the downstream vision task, performing a forward pass execution of each model to generate a loss, backpropagating the loss to modify parameter values, and iterating the forward pass execution and backpropagating with images generated by the modified parameters to jointly train the models and optimize the parameters.

Abstract: An image processing apparatus includes a representative-distance calculator and a joining processor. The representative-distance calculator generates representative distance values of the basis of a distance image generated from a stereo image including an image of at least one object including a vehicle. The distance image includes distance values of pixels. The joining processor performs a joining process of joining a first image region and a second image region that are defined based on the image of the at least one object and disposed apart from each other in the stereo image. The joining processor performs a determining process of determining whether the first image region and the second image region each include an image of a side face of the vehicle. The joining processor performs the joining process when the first image region and the second image region each include the image of the side face of the vehicle.

Abstract: Systems and methods are described for the compositing together of model-linked vascular data from a plurality of sources, including at least one 2-D angiography image, for display in a frame of reference of the at least one angiography image. In some embodiments, a linking model comprises a data structure configured to link locations of angiographic images to corresponding elements of non-image vascular parameter data. The linking data structure is traversed to obtain a mapping to the frame of reference of one or more of the angiographic images.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium for red eye detection are provided. In one aspect, a system includes an image acquisition device, first and second illuminators, and at least one processor. The first illuminator is arranged closer to the image acquisition device than the second illuminator. The image acquisition device is configured to capture a first facial image of a face of a subject with the first illuminator being on and the second illuminator being off and a second facial image of the face of the subject with the second illuminator being on and the first illuminator being off. The processor can process the first facial image based on the second facial image to determine whether at least one eye of the subject is live by determining that the first facial image includes a red eye reflection from the at least one eye.

Abstract: A division line detection device includes: a processor configured to detect a division line candidate pixel from an image acquired by a camera mounted on a vehicle, set first reliability for a division line candidate pixel whose value representing likelihood that a lane division line is represented is equal to or more than a predetermined threshold value; set second reliability lower than the first reliability for a division line candidate pixel whose value is less than the predetermined threshold value; correct to the first reliability, when a first predetermined number or more of the division line candidate pixels are located on a first scan line having one end at a vanishing point of the image, the second reliability of each division line candidate pixel on the first scan line; and detect a lane division line based on the division line candidate pixels having the first reliability.

Abstract: A method for predicting the realism of an object within an image includes generating a training image set for a predetermined object type. The training image set comprises one or more training images at least partially generated using a computer. A pixel level training spatial realism map is generated for each training image of the one or more training images. Each training spatial realism map configured to represent a perceptual realism of the corresponding training image. A predictor is trained using the training image set and the corresponding training spatial realism maps. An image of the predetermined object is received. A spatial realism map of the received image is produced using the trained predictor.

Abstract: An evaluation system that may include an imager; and a processing circuit. The imager may be configured to obtain an electron image of a hole that is formed by an etch process, the hole exposes at least one layer of a one or more sets of layers, each set of layers comprises layers that differ from each other by their electron yield and belong to an intermediate product. The processing circuit may be configured to evaluate, based on the electron image, whether the hole ended at a target layer of the intermediate product. The intermediate product is manufactured by one or more manufacturing stages of a manufacturing process of a three dimensional NAND memory unit. The hole may exhibit a high aspect ratio, and has a width of a nanometric scale.

Abstract: Systems and methods for analyzing image data to automatically ensure image capture consistency are described. According to certain aspects, a server may access aerial and other image data and identify a set of parameters associated with the capture of the image data. The server may access a corresponding set of acceptable image capture parameters, and may compare the set of parameters to the set of acceptable parameters to determine whether image data is consistent with the set of acceptable parameters. In some embodiments, if the image data is not consistent, the server may generate a notification or instruction to cause an image capture component to recapture additional image data.

Abstract: The present disclosure relates to a target object detection method, a readable storage medium and an electronic device. The method is periodically executed and the method includes: identifying a target object based on first N frames of image data obtained within a current execution period, wherein N is a positive integer, and N is less than the total number of frames of image data that can be obtained within the current execution period; and if the target object is identified, tracking the target object based on the obtained image data within the remaining time period of the current execution period. In the above technical solution, when the target object is identified, the identified target object is tracked within the remaining time period of the current execution period, so that the data calculation amount can be effectively reduced.

Abstract: A depth detection device includes a stereo image recording device, a scaler and a depth determiner is provided. The stereo image recording device is configured to record a stereo image of a scene, and includes a first optical path for recording a first image of the stereo image, and a second optical path for recording a second image of the stereo image. The first optical path and the second optical path have different lengths. The scaler determines a scaling of the first image or of the second image, by minimizing differing lateral magnification of objects in the scene, and for performing the scaling. The depth determiner determines a depth map of the scene based upon the scaled stereo image.