Abstract: Methods, devices, and computer-readable media for multi-task based lifelong learning. A method for lifelong learning includes identifying a new task for a machine learning model to perform. The machine learning model trained to perform an existing task. The method includes adaptively training a network architecture of the machine learning model to generate an adapted machine learning model based on incorporating inherent correlations between the new task and the existing task. The method further includes using the adapted machine learning model to perform both the existing task and the new task.

Abstract: Techniques are disclosed for determining whether to include a bodyprint in a cluster of bodyprints associated with a recognized person. For example, a device performs facial recognition to identify the identity of a first person. The device also identifies and stores physical characteristic information of the first person, the stored information associated with the identity of the first person based on the recognized face. Subsequently, the device receives a second video feed showing an image of a second person whose face is also determined to be recognized by the device. The device then generates a quality score for physical characteristics in the image of the user. The device can then add the image with the physical characteristics to a cluster of images associated with the person if the quality score is above a threshold, or discard the image if not.

Abstract: Techniques are disclosed for providing a notification indicating an identity of a first person based on face-associated body characteristics. For example, a device performs facial recognition to identify the identity of the first person shown in a first video feed. The device also identifies and stores physical characteristic information of the first person from the first video feed, the stored information associated with the identity of the first person based on the recognized face. Subsequently, the device receives a second video feed showing a second person whose face is determined to not be recognized by the device. The device compares the stored physical characteristic information of the first person with additional physical characteristic information of the second person shown in the second video feed. Based on the comparison, the device provides a notification indicating whether the identity of the second person corresponds to the identity of the first person.

Abstract: Techniques are disclosed for determining the presence of a particular person based on facial characteristics. For example, a device may include a first image in a reference set of images based on determining that a face shown in the first image is not covered by a face covering. A trained model of the device may determine a first set of characteristics from the first image, whereby the trained model is trained utilizing simulated face coverings to match a partially covered face of a particular person with a non-covered face of the particular person. The device may also determine a second set of characteristics associated with a face of a second person based on a second image. The trained model may then determine a score corresponding to a level of similarity between both sets of characteristics, and then determine whether the first person is the second person based on the score.

Abstract: The present application provides a near field communication antenna structure, a housing with the same, and an electronic terminal, the near field communication antenna structure including a first conductive body, a second conductive body, a micro-slit structure, a first feed point and a second feed point, wherein the micro-slit structure is positioned between the first conductive body and the second conductive body to separate the first conductive body from the second conductive body, the first feed point is provided on the first conductive body, and the second feed point is provided on the second conductive body. The use of the above NFC antenna structure saves a separate NFC antenna structure, reduces cost and has a simple structure.

Abstract: One embodiment provides a method comprising identifying a salient part of an object in an input image based on processing of a region of interest (RoI) in the input image at an electronic device. The method further comprises determining an estimated full appearance of the object in the input image based on the salient part and a relationship between the salient part and the object. The electronic device is operated based on the estimated full appearance of the object.

Abstract: Embodiments are disclosed of a radiator temperature control apparatus for controlling the heat output of a radiator. The radiator temperature control apparatus may include an airtight enclosure around the air outlet of the radiator air vent, an adjustable opening in the airtight enclosure controlled by an actuator, and a controller connected to the actuator. In operation, the controller can be configured to open the adjustable opening in the airtight enclosure allowing air in the radiator to be expelled through the adjustable opening, thereby allowing steam to enter the radiator, and heat the room. The controller can be configured to close the adjustable opening, stopping air from being expelled from the radiator, thereby stopping additional steam from entering the radiator.

Abstract: A method, a computer readable medium, and a system for cell annotation are disclosed. The method includes receiving at least one new cell image for cell detection; extracting cell features from the at least one new cell image; comparing the extracted cell features to a matrix of cell features of each class to predict a closest class, wherein the matrix of cell features has been generated from at least initial training data comprising at least one cell image; detecting cell pixels from the extracted cell features of the at least one new cell image using the predicted closest class to generate a likelihood map; extracting individual cells from the at least one cell image by segmenting the individual cells from the likelihood map; and performing a machine annotation on the extracted individual cells from the at least one new cell image to identify cells, non-cell pixels, and/or cell boundaries.

Abstract: An artificial neural network system for image classification, formed of multiple independent individual convolutional neural networks (CNNs), each CNN being configured to process an input image patch to calculate a classification for the center pixel of the patch. The multiple CNNs have different receptive field of views for processing image patches of different sizes centered at the same pixel. A final classification for the center pixel is calculated by combining the classification results from the multiple CNNs. An image patch generator is provided to generate the multiple input image patches of different sizes by cropping them from the original input image. The multiple CNNs have similar configurations, and when training the artificial neural network system, one CNN is trained first, and the learned parameters are transferred to another CNN as initial parameters and the other CNN is further trained. The classification includes three classes, namely background, foreground, and edge.

Abstract: Methods, devices, and computer-readable media for multi-task based lifelong learning. A method for lifelong learning includes identifying a new task for a machine learning model to perform. The machine learning model trained to perform an existing task. The method includes adaptively training a network architecture of the machine learning model to generate an adapted machine learning model based on incorporating inherent correlations between the new task and the existing task. The method further includes using the adapted machine learning model to perform both the existing task and the new task.

Abstract: One embodiment provides a method comprising identifying a salient part of an object in an input image based on processing of a region of interest (RoI) in the input image at an electronic device. The method further comprises determining an estimated full appearance of the object in the input image based on the salient part and a relationship between the salient part and the object. The electronic device is operated based on the estimated full appearance of the object.

Abstract: An artificial neural network system for image classification, formed of multiple independent individual convolutional neural networks (CNNs), each CNN being configured to process an input image patch to calculate a classification for the center pixel of the patch. The multiple CNNs have different receptive field of views for processing image patches of different sizes centered at the same pixel. A final classification for the center pixel is calculated by combining the classification results from the multiple CNNs. An image patch generator is provided to generate the multiple input image patches of different sizes by cropping them from the original input image. The multiple CNNs have similar configurations, and when training the artificial neural network system, one CNN is trained first, and the learned parameters are transferred to another CNN as initial parameters and the other CNN is further trained. The classification includes three classes, namely background, foreground, and edge.

Abstract: An artificial neural network system for image classification, including multiple independent individual convolutional neural networks (CNNs) connected in multiple stages, each CNN configured to process an input image to calculate a pixelwise classification. The output of an earlier stage CNN, which is a class score image having identical height and width as its input image and a depth of N representing the probabilities of each pixel of the input image belonging to each of N classes, is input into the next stage CNN as input image. When training the network system, the first stage CNN is trained using first training images and corresponding label data; then second training images are forward propagated by the trained first stage CNN to generate corresponding class score images, which are used along with label data corresponding to the second training images to train the second stage CNN.

Abstract: Pathological analysis needs instance-level labeling on a histologic image with high accurate boundaries required. To this end, embodiments of the present invention provide a deep model that employs the DeepLab basis and the multi-layer deconvolution network basis in a unified model. The model is a deeply supervised network that allows to represent multi-scale and multi-level features. It achieved segmentation on the benchmark dataset at a level of accuracy which is significantly beyond all top ranking methods in the 2015 MICCAI Gland Segmentation Challenge. Moreover, the overall performance of the model surpasses the state-of-the-art Deep Multi-channel Neural Networks published most recently, and the model is structurally much simpler, more computational efficient and weight-lighted to learn.

Abstract: A method, a computer readable medium, and a system for cell annotation are disclosed. The method includes receiving at least one new cell image for cell detection; extracting cell features from the at least one new cell image; comparing the extracted cell features to a matrix of cell features of each class to predict a closest class, wherein the matrix of cell features has been generated from at least initial training data comprising at least one cell image; detecting cell pixels from the extracted cell features of the at least one new cell image using the predicted closest class to generate a likelihood map; extracting individual cells from the at least one cell image by segmenting the individual cells from the likelihood map; and performing a machine annotation on the extracted individual cells from the at least one new cell image to identify cells, non-cell pixels, and/or cell boundaries.

Abstract: A method, computer readable medium, and system are disclosed of enhancing cell images for analysis. The method includes performing a multi-thresholding process on a cell image to generate a plurality of images of the cell image; smoothing each component within each of the plurality of images; merging the smoothed components into a merger layer; classifying each of the components of the merged layer into convex cell regions and concave cell regions; combining the concave cell regions with a cell boundary for each of the corresponding concave cell regions to generate a smoothed shape profile for each of the concave cell regions; and generating an output image by combining the convex cell regions with the concave cell regions with smoothed shape profiles.

Abstract: Semiconductor packages and methods of manufacturing semiconductor packages are described herein. In certain embodiments, the semiconductor package includes a substrate, a wall attached to the substrate, a first adhesive layer disposed between a bottom surface of the wall and a top surface of the substrate, and a second adhesive layer disposed around an outer perimeter of the first adhesive layer, the second adhesive layer disposed adjacent and contacting the wall, the second adhesive layer different from the first adhesive layer, wherein at least one of the first adhesive layer and the second adhesive layer connects the wall to electrical ground.

Abstract: A method and system are disclosed for recognizing an object, the method including emitting one or more arranged patterns of infrared rays (IR) from an infrared emitter towards a projection region, the one or more arranged patterns of infrared rays forming unique dot patterns; mapping the one or more arranged patterns of infrared rays on the operation region to generate a reference image; capturing an IR image and a RGB image of an object with a wearable device, the wearable device including an infrared (IR) camera and a RGB camera; extracting IR dots from the IR image and determining a match between the extracted IR dots and the reference image; determining a position of the RGB image on the reference image; and mapping the position of the RGB image to a coordinate on the projection region.

Abstract: Semiconductor packages and methods of manufacturing semiconductor packages are described herein. In certain embodiments, the semiconductor package includes a substrate, a wall attached to the substrate, a first adhesive layer disposed between a bottom surface of the wall and a top surface of the substrate, and a second adhesive layer disposed around an outer perimeter of the first adhesive layer, the second adhesive layer disposed adjacent and contacting the wall, the second adhesive layer different from the first adhesive layer, wherein at least one of the first adhesive layer and the second adhesive layer connects the wall to electrical ground.