Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.

Abstract: Methods, systems, and computer programs are presented for detecting near-duplicate profile images of the users in a social network. One method includes operations for identifying an image in a profile of a user of the social network, determining a query feature vector for the image, the query feature vector comprising a set of features, and determining a dominant feature from the features, the dominant feature having a highest value from the values of the features. Further, the method includes operations for determining a bucket in a database of feature vectors based on the dominant feature, determining if the query feature vector is a near duplicate of any feature vector in the determined bucket, and determining if the profile of the user is a duplicate profile or a fake profile based on whether the query feature vector is a near duplicate of any feature vector in the determined bucket.

Abstract: In an optical character recognition (OCR) method for digitizing printed text images using a long-short term memory (LSTM) network, text images are pre-processed using a stroke-aware max-min pooling method before being fed into the network, for both network training and OCR prediction. During training, an average stroke thickness is computed from the training dataset. Stroke-aware max-min pooling is applied to each text line image, where minimum pooling is applied if the stroke thickness of the line is greater than the average stroke thickness, while max pooling is applied if the stroke thickness is less than or equal to the average stroke thickness. The pooled images are used for network training. During prediction, stroke-aware max-min pooling is applied to each input text line image, and the pooled image is fed to the trained LSTM network to perform character recognition.

Abstract: Various embodiments include systems and methods that provide a socially interactive application for requesting, searching, and sharing visual step-by-step instructions and related information to complete a project.

Abstract: In an optical character recognition (OCR) method for digitizing printed text images using a long-short term memory (LSTM) network, text images are pre-processed using a stroke-aware max-min pooling method before being fed into the network, for both network training and OCR prediction. During training, an average stroke thickness is computed from the training dataset. Stroke-aware max-min pooling is applied to each text line image, where minimum pooling is applied if the stroke thickness of the line is greater than the average stroke thickness, while max pooling is applied if the stroke thickness is less than or equal to the average stroke thickness. The pooled images are used for network training. During prediction, stroke-aware max-min pooling is applied to each input text line image, and the pooled image is fed to the trained LSTM network to perform character recognition.

Abstract: Methods and systems for monitoring process equipment such as field devices. A QR code can be associated with a field device, wherein the QR code contains data that identifies the field device, and also includes process data regarding the field device, the location of the field device, and maintenance information, installation information and fault information associated with the field device. The QR code can then be scanned and decoded in order to retrieve the data for use in in monitoring and maintaining field devices in the context of a connected plant.

Abstract: Optical wireless communications for vehicle operator personal computing devices is provided. In one embodiment, a system comprises: a bus I/O processor coupled to a vehicle data bus, the I/O processor executing a protocol converter; an optical communication transceiver coupled to the I/O processor, the transceiver in communication with an optical unit comprising an optical emitter and sensor, the optical unit within a control compartment of the vehicle; the bus I/O processor exchanges vehicle operational data with on-board data processing devices over the vehicle data bus using a vehicle data bus protocol; the transceiver controls the optical unit to establish a bidirectional optical wireless communications (OWC) link with a personal computing device in the control compartment using an OWC protocol; the I/O processor communicates information relevant to the operation of the vehicle between an on-board data processing device and an application executing on the personal computing device over the OWC link.

Abstract: In a text line segmentation process, connected components (CCs) in document image are categorized into three subsets (normal, large, small) based on their sizes. The centroids of the normal size CCs are used to perform line detection using Hough transform. Among the detected candidate lines, those with line bounding box heights greater than a certain height are removed. For each normal size CC, if its bounding box does not overlap the bounting box of any line with an overlap area greater than a predefined fraction of the CC bounding box, a new line is added for this CC, which passes through the centroid of the CC and has an average slant angle. Each large size CCs are broken into two or more CCs. All CCs are then assigned to the nearest lines. A refinement method is also described, which can take any text line segmentation result and refine it.

Abstract: In a text line segmentation process, connected components (CCs) in document image are categorized into three subsets (normal, large, small) based on their sizes. The centroids of the normal size CCs are used to perform line detection using Hough transform. Among the detected candidate lines, those with line bounding box heights greater than a certain height are removed. For each normal size CC, if its bounding box does not overlap the bounting box of any line with an overlap area greater than a predefined fraction of the CC bounding box, a new line is added for this CC, which passes through the centroid of the CC and has an average slant angle. Each large size CCs are broken into two or more CCs. All CCs are then assigned to the nearest lines. A refinement method is also described, which can take any text line segmentation result and refine it.

Abstract: An artificial neural network system implemented on a computer for cell segmentation and classification of biological images. It includes a deep convolutional neural network as a feature extraction network, a first branch network connected to the feature extraction network to perform cell segmentation, and a second branch network connected to the feature extraction network to perform cell classification using the cell segmentation map generated by the first branch network. The feature extraction network is a modified VGG network where each convolutional layer uses multiple kernels of different sizes. The second branch network takes feature maps from two levels of the feature extraction network, and has multiple fully connected layers to independently process multiple cropped patches of the feature maps, the cropped patches being located at a centered and multiple shifted positions relative to the cell being classified; a voting method is used to determine the final cell classification.

Abstract: A local connectivity feature transform (LCFT) is applied to binary document images containing text characters, to generate transformed document images which are then input into a bi-directional Long Short Term Memory (LSTM) neural network to perform character/word recognition. The LCFT transformed image is a gray scale image where the pixel values encode local pixel connectivity information of corresponding pixels in the original binary image. The transform is one that provides a unique transform score for every possible shape represented as a 3×3 block. In one example, the transform is computed using a 3×3 weight matrix that combines bit coding with a zigzag pattern to assign weights to each element of the 3×3 block, and by summing up the weights for the non-zero elements of the 3×3 block shape.

Abstract: An artificial neural network system implemented on a computer for cell segmentation and classification of biological images. It includes a deep convolutional neural network as a feature extraction network, a first branch network connected to the feature extraction network to perform cell segmentation, and a second branch network connected to the feature extraction network to perform cell classification using the cell segmentation map generated by the first branch network. The feature extraction network is a modified VGG network where each convolutional layer uses multiple kernels of different sizes. The second branch network takes feature maps from two levels of the feature extraction network, and has multiple fully connected layers to independently process multiple cropped patches of the feature maps, the cropped patches being located at a centered and multiple shifted positions relative to the cell being classified; a voting method is used to determine the final cell classification.

Abstract: Methods and systems for identifying and connecting to field devices in an industrial plant. In an example embodiment, steps or operations can be implemented for providing credentials associated with a user through Li-Fi wireless communications, detecting a light pattern associated with the Li-Fi Wireless communications with a light sensor attached to a field device, transmitting a signal from the light sensor to a controller, which decodes the credentials associated with the user, and if the credentials are authorized, allowing the user to obtain device data associated with the field device wherein said data includes a location of said field device within said industrial plant. Such data can be obtained from a control room and/or a server through a wireless data communications network (e.g., wireless bidirectional communications).

Abstract: Methods and systems for monitoring process equipment such as field devices. A QR code can be associated with a field device, wherein the QR code contains data that identifies the field device, and also includes process data regarding the field device, the location of the field device, and maintenance information, installation information and fault information associated with the field device. The QR code can then be scanned and decoded in order to retrieve the data for use in in monitoring and maintaining field devices in the context of a connected plant.

Abstract: Optical wireless communications for vehicle operator personal computing devices is provided. In one embodiment, a system comprises: a bus I/O processor coupled to a vehicle data bus, the I/O processor executing a protocol converter; an optical communication transceiver coupled to the I/O processor, the transceiver in communication with an optical unit comprising an optical emitter and sensor, the optical unit within a control compartment of the vehicle; the bus I/O processor exchanges vehicle operational data with on-board data processing devices over the vehicle data bus using a vehicle data bus protocol; the transceiver controls the optical unit to establish a bidirectional optical wireless communications (OWC) link with a personal computing device in the control compartment using an OWC protocol; the I/O processor communicates information relevant to the operation of the vehicle between an on-board data processing device and an application executing on the personal computing device over the OWC link.

Abstract: A method of providing purchase recommendations to a user may include tracking user comparisons of products on an electronic commerce system, receiving a user selection of an anchor product from the products through an electronic user interface of the electronic commerce system, designating a recommended product from the products for recommendation to the user through the electronic commerce system according to a frequency with which the recommended product is compared with the anchor product based on the tracking, and presenting the designated recommended product to the user responsive to the user's selection of the anchor product.

Abstract: This disclosure includes systems and methods for providing purchase recommendations to a user that may include items frequently purchased with a product selected by the user. The determination of which items are frequently purchased with which other items may account for both online and in-store transactions and may further account for both pairwise and multi-wise relationships. The recommendations may be provided on an electronic user interface, such as a website, in response to the user's selection of the product through the electronic user interface. The recommendations may be tailored to the user's selected product so that the recommended items are available in the same delivery channel as the user-selected product.

Abstract: Methods and apparatus are disclosed provide various enhancements and/or improvements to the conventional shopping process used by brick-and-mortar stores. In some embodiments, a customer may use a mobile computing device, such as a smart phone, to select and pay for items in a brick-and-mortar store. Such shopping process may permit the customer to purchase numerous items without the using of shopping basket or a wheeled shopping cart. The shopping process may further permit the customer to pay for items and leave the store with such items without waiting in a checkout line.

Abstract: A system and method supporting the assisted creation of media content documenting the details of completing a step-by-step project by a consumer in an Internet-based, social networking/social e-commerce environment.

Abstract: A computing device (e.g., a server) coupled to a digital signage device may receive characteristic information associated with a visitor of the digital signage device. The received characteristic information may be analyzed, and based on a result of the analysis, advertisement attributes may be determined. The computing device may select, based on the determined advertisement attributes, a particular advertisement from a plurality of advertisement bids. The computing device may cause display of the selected particular advertisement on the digital signage device for the visitor. The characteristic information may be generated by the digital signage device utilizing a video analytics. The computing device may receive particular gesture information associated with a mobile device of the visitor who responds to the displayed particular advertisement.