Abstract: A set of enhancements to further improve the performance of deep learning artificial intelligence algorithms trained to detect and localize colon polyps. The enhancements spanning training data mining efficiencies and automation, training data augmentation, early detection of polyps enable a more performant colon polyp detection solution for use on colonoscopy procedure recordings or live procedures in endoscopy centers.

Abstract: An example apparatus has a processor to analyze images of projected shifted versions of a fringe pattern onto a scene to obtain phase information associated with the pixels in the image. Topology information is derived by correcting the phase information using a phase offset associated with a combination of two subsequent versions of the fringe pattern in the images or by estimating a surface normal for each pixel using a partial derivative of the phase information of the pixel in a first spatial direction and a partial derivative of the phase information of the pixel in a second spatial direction.

Abstract: A user device may capture a plurality of preview images that each include image data. The user device may process, in substantially real-time, a preview image to identify an object in the preview image, including determining an outline of the object. The user device may create a mask that hides a portion of the image data associated with an area outside of the outline of the object in one or more of the preview images. Based on determining that a parameter associated with glare of the object in the preview images does not satisfy a threshold, the user device may provide, in substantially real-time, feedback to a user, including an instruction to the user to perform an action with respect to the user device or to the object, and, based on determining that the parameter satisfies the threshold, the user device may automatically capture an image of the object.

Abstract: A video coder that implicitly signals a transform setting for coding a block of pixels is provided. The video coder derives a transform setting for a block of pixels based on a block processing setting. The video coder processes the block of pixels according to the block processing setting. For encoding, the video coder transforms a set of residual pixels to generate a set of transform coefficients according to the transform setting. For decoding, the video coder inverse transforms the transform coefficients to generate a set of residual pixels according to the transform setting.

Abstract: A mobile robot including an image pickup unit, further includes an accumulation unit configured to accumulate imaging data taken in the past, a reception unit configured to receive designation of a spatial area from a remote terminal, and a processing unit configured, when it is possible to shoot the spatial area received by the reception unit by the image pickup unit, to perform shooting and transmit obtained imaging data to the remote terminal, and when the shooting is impossible, to transmit imaging data including an image of the spatial area accumulated in the accumulation unit to the remote terminal and start a moving process in order to shoot the spatial area by the image pickup unit.

Abstract: Video coding techniques including differential bit rate or quality coding of one or more regions of interest and one or more non-regions of interest based on information including one or more of coordinates of the one or more regions of interest, a target complexity, residual encoder bit data, a requested quality, a difference between the current video data frame and a reconstructed video data frame, a target quality, a requested bit rate, frame target bit allocation and an as encoded bit rate.

Abstract: A volume measuring apparatus having a first camera, a second camera, an emitting unit and a processing unit is disclosed. The processing unit controls the emitting unit to emit invisible structure light, and controls the first and second camera to capture a left and a right image both containing a target-box. The processing unit generates a depth graph according to the left and right image, and scans the depth graph through multiple scanning lines for determining a middle line, a bottom line, a left-sideline, and a right-sideline of the target-box in the depth graph. The processing unit performs scanning, within a range of the middle line, the bottom line, the left-sideline, and the right-sideline, for obtaining a plurality of width information, height information, and length information. The processing unit computes the volume related data of the target-box according to the plurality of width information, height information, and length information.

Abstract: A method and device for encoding an image divided into blocks. The image contains two separate zones. The method implements the following: only in the event where the current block pertains to one of the zones of the image: encoding the current block using a prediction, the current block being predicted using a previously encoded and then decoded block located in the other zone of the image, the blocks of the other zone having been previously encoded and then decoded, and encoding information indicating the application of the prediction; and for any decoded block pertaining to the other zone of the image, storing the decoded block data.

Abstract: A 360-degree image encoding method that is performed by an encoding apparatus according to the present disclosure comprises the steps of: obtaining a two-dimensional (2D) space picture with respect to 360-degree video data; deriving motion constrained tile sets (MCTSs) for configuring a sub-picture from the 2D picture; configuring the sub-picture including the MCTSs based on the MCTSs; and encoding the sub-picture so as to output encoded image information, wherein the MCTSs are arranged in the raster scan order on the sub-picture.

Abstract: One variation of a method for modeling a human body includes: driving a sensor block along a path above a platform occupied by a user and recording a sequence of depth images and color images, via the sensor block, at each capture position in a sequence of capture positions along the path; compiling the set of depth images into a low-density 3D model of the user and a high-density 3D model of the user; extracting a set of color patches, from the sequence of color images, corresponding to discrete regions on a surface of the low-density 3D model of the user; projecting the set of color patches onto the surface of the low-density 3D model to generate a 3D color model of the user; and extracting a value of a dimension, projected from the low-density 3D color model onto the high-density 3D model, from the high-density 3D model.

Abstract: A method and apparatus for performing group marking on a uniform transform unit mode and an electronic device are provided. The method includes: marking group units according to uniform transform unit modes of coding units included in the group units; wherein when the uniform transform unit modes of the coding units in the group units have a first value, a flag is set to be of a second value, and the group units are marked by using the first flag; and marking the coding units in the group units according to the uniform transform unit modes of the coding units included in the group units. According to this disclosure, when the uniform transform unit modes of the coding units in the group units are identical, the uniform transform unit modes of the coding units in the group units are not marked, thereby saving storage capability, transmission capability and processing capability.

Abstract: A hand-held image-capturing electronic device with ability to compensate for unstable rotation and images during 360-degree panoramic captures includes a display screen, a first lens unit, a second lens unit, and a third lens unit. The first lens unit and the second lens unit are positioned on opposing surfaces, the third lens unit is independently rotatable on the electronic device and can cooperate with the first lens unit and the second lens unit to capture images which are refined and synthesized together by the device. A method for capturing such images with such device is also disclosed.

Abstract: An endoscopy video feature enhancement platform (EVFEP) is connected to the output of any type endoscope system, and inputs and captures the output video. The video is visually augmented live with indicators of possible polyp detection and localization, polyp attributes, and procedure metrics, based on the collective learning of the output results of many different types of endoscopy systems on a large scale. An artificial intelligence model is trained on confirmed polyp detection previously determined by this and other EVFEP devices used with many different types of endoscope systems on a large scale. Augmented video, images and automatically generated short video clips of key procedure segments are passed to a reporting system, and supplemented with meta data.

Abstract: An inter prediction method according to the present invention comprises: a step for deriving reference motion information related to a unit to be decoded in a current picture; and a step for performing motion compensation for the unit to be decoded, using the reference motion information that has been derived. According to the present invention, image encoding/decoding efficiency can be enhanced.

Abstract: A system and method for improving quality of encoding and decoding a panoramic video. The panoramic video comprises a sequence of encoded picture frames. In panoramic video picture frames are mapped on a spherical viewing area. In the method the projection center of the encoded frame is transformed before encoding. In the decoding phase the projection center is transformed back accordingly after decoding the frame.

Abstract: The present disclosure concerns a method and a device for encoding or decoding video data. It concerns more particularly the encoding according to a particular encoding mode using a decoder side motion vector derivation mode referenced as frame-rate up conversion mode or FRUC mode. It concerns encoding and decoding improvement which reduce the need for memory accesses when using an encoding mode where the motion information is predicted using a decoder side motion vector derivation method.

Abstract: A system for downhole optical imaging includes a housing forming at least a portion of a tool string. The system also includes a source arranged within the housing, the source emitting light through a window formed in the housing. The system further includes an imager arranged within the housing, the imager receiving imaging data through the window. The system also includes a control system communicatively coupled to the imager, the control system processing the image data using one or more algorithms, the one or more algorithms modifying the imaging data based at least in part on a scattering material surrounding the housing.

Abstract: A method and apparatus of video coding incorporating Deep Neural Network are disclosed. A target signal is processed using DNN (Deep Neural Network), where the target signal provided to DNN input corresponds to the reconstructed residual, output from the prediction process, the reconstruction process, one or more filtering processes, or a combination of them. The output data from DNN output is provided for the encoding process or the decoding process. The DNN can be used to restore pixel values of the target signal or to predict a sign of one or more residual pixels between the target signal and an original signal. An absolute value of one or more residual pixels can be signalled in the video bitstream and used with the sign to reduce residual error of the target signal.

Abstract: A system for verifying medication doses in a filled medication package comprises a camera(s), a contour light source(s), a set of relief light sources. A verification unit may be used for imaging a contour of the medication items, imaging a surface relief of the medication items, processing the images, confirming the content of the medication package relative to identification of a type of medication item and/or medication package using the processing of the images. An interface produces a verification output based on the confirmation by the verification unit.

Abstract: A photo-optic comparative geolocation system for calculating the location of an object has been developed. The apparatus includes optic sensors that capture surrounding location data, an interface that maps the optic sensor data, a storage database containing prior optical and location data, a digital sextant that provides data calculated using magnetic or celestial references, a data processor that compares the mapped data to stored data and calculates current location based on the comparison analysis, and a visual display for location information.