A method of recalibrating a feature data of each channel generated by a convolution layer of a convolution neural network is provided. According to some embodiments, since an affine transformation is applied to the feature data of each channel independently of the feature data of the other channel, the overall number of parameters defining the affine transformation is minimized. As a result, the amount of computations required in performing the feature data recalibration can be reduced.

A system and method for performing real-time quality inspection of objects is disclosed. The system and method include a transport to move objects being inspected, allowing the inspection to be performed in-line. At least one optical acquisition unit is provided that captured optical images of the objects being inspected. The captured optical images are matched to CAD models of objects, and the matched CAD model is extracted. A laser with an illumination light beam has a wavelength in the violet or ultraviolet range then conducts scans of the objects, which are formed into three-dimensional point clouds. The point clouds are compared to the extracted CAD models for each object, where CTF are compared to user input or CAD model information and the object is determined to be acceptable or defective based on the extent of deviation between the point cloud and the CAD model.

A method and a client are provided. The method includes determining a first parameter according to a detected first operation, the first parameter being used for indicating a cut-out area. A first image corresponding to the cut-out area is obtained according to the first parameter. Recognition is performed on the first image, and based on a recognition result, recommendation information associated with the first image is obtained. A second operation of selecting target information from the recommendation information is detected, and the selected target information is added to the first image to generate a second image.

This application discloses a repositioning method performed by an electronic device in a camera pose tracking process, belonging to the field of augmented reality (AR). The method includes: obtaining a current image acquired by the camera after an i.sup.th anchor image in a plurality of anchor images; obtaining an initial feature point and an initial pose parameter in a first anchor image in the plurality of anchor images in a case that the current image satisfies a repositioning condition; performing feature point tracking on the current image relative to the first anchor image, to obtain a target feature point; calculating a pose change amount of a camera from a first camera pose to a target camera pose according to the initial feature point and the target feature point; and performing repositioning according to the initial pose parameter and the pose change amount to obtain a target pose parameter.

A method implemented on a computing device comprising a data-parallel coprocessor and a memory coupled with the data-parallel processor for generating and evaluating N-to-1 mappings between spatial point sets in nD. The method comprises using the computing device to carry out steps comprising receiving a first and a second spatial point sets, an array of (n+1) combinations in the first spatial point set, an array of one or more pairs of neighbor (n+1) combinations referencing into the array of (n+1) combinations, and a CCISS between the two spatial point sets; computing a plurality of solving structures and provide a two-level indexing structure for (n+1) combinations for the plurality of (n+1) combinations; generating one or more N-to-1 mappings; and generating a plurality of local distance measures for unique combinations of the one or more pairs of neighbor (n+1) combinations and the one or more N-to-1 mappings. Some embodiments further comprises providing in addition a two-level indexing structure for pairs of neighbor (n+1) combinations for generating the plurality of local distance measures.

A computer-based system (e.g., a handheld device) is configured to detect, count, and measure logs in a stack. A user captures one or more images of the stack. Where multiple images are captured, the system creates a working image by stitching together portions of the image. The system identifies a contour indicating the outline of the stack in the working image and fits ellipses to the log faces in the working image. Information such as the number of logs, the volume of wood in the stack, and the average log diameter may be made available for presentation to the user.

A microscopy system includes a first excitation source configured to emit a one-photon light beam and a second excitation source configured to emit a two-photon light beam. The system also includes a first optical component configured to deliver one or more of a one-photon light sheet and a two-photon light sheet, where the one-photon light sheet is based on the one-photon light beam and the two-photon light sheet is based on the two-photon light beam. The system further includes a single reflective surface at a back focal plane of the first optical component to reflect the one or more of the one-photon light sheet and the two-photon light sheet toward a sample. The single reflective surface rotates about a fixed axis situated at a cross section of the single reflective surface and the back focal plane of the first optical component.

Systems and methods for encoding multiple video streams in accordance with embodiments of the invention are disclosed. In one embodiment, a source encoder configured to encode source video as a number of alternative video streams includes a parallel processing system configured to collect statistics on source video data and write the statistics to shared memory in a first pass through the received multimedia content, determine initial encoding information for source video data and write the initial encoding information to shared memory during the first, encode the source video data in parallel using collected statistics and initial encoding information to produce a plurality of alternative video streams during a second pass, and parallel encoding processes are configured to reuse additional encoding information that has already been determined for a portion of video and to generate additional encoding information that has not already been determined for a portion of video.

The present invention provides systems and methods for tracking items (e.g., commodities, goods, containers, boxes, packages, etc.) through transportations to multiple locations within a pre-defined space, to allow the position(s) and movement(s) of such items to be accurately tracked and documented, and to allow such items to be quickly identified and located based on tracking records kept within the tracking system. The system may utilize image sensors, image recognition and processes software, position translation software, and a virtual model of the pre-defined space in order to track objects within the defined space and maintain a record of the movement(s) and position(s) of the object within the pre-defined space.

A system processes images of documents, for example, identification documents. The system transforms an image of a document to generate an image that represent the document in a canonical form. For example, if the input image has a document that is tilted at an angle with respect to the sides of the image, the system modifies the orientation of the document to show the document having sides aligned with the sides of the image. The system stores user accounts that include user information including images. The system generates a graph of nodes that represent user accounts with edges determined based on similarity scores between user accounts. The system determines connected components of user accounts, such that each connected component represents user accounts that have a high likelihood of being duplicates.