An intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact.

Intelligent Automated Assistant

A mobile terminal including a body; a touchscreen provided to a front and extending to side of the body and configured to display content; and a controller configured to detect one side of the body comes into contact with one side of an external terminal, display a first area on the touchscreen corresponding to a contact area of the body and the external terminal and a second area including the content, receive an input of moving the content displayed in the second area to the first area, display the content in the first area, and share the content in the first area with the external terminal.

Mobile terminal and method for controlling the same

A projection exposure apparatus including an irradiation optical system including a light source and irradiating a mask with irradiation light beams, a projection optical system for projecting an image of a pattern of the mask on a substrate, a plurality of first fly-eye type optical integrators each having an emission side focal plane disposed on a Fourier transformed surface with respect to the pattern of the mask in the irradiation optical system or on a plane adjacent to the same and having a center located at a plurality of positions which are eccentric from the optical axis of the irradiation optical system, a plurality of second fly-eye type optical integrators each having an emission side focal plane disposed on a Fourier transformed plane with respect to the incidental end of each of a plurality of the first fly-eye type optical integrators or on a plane adjacent to the same and being disposed to correspond to a plurality of the first fly-eye type optical integrators, and a light divider for dividing and causing the irradiation light beams from the light source to be incident on each of a plurality of the second fly-eye type optical integrators.

Projection exposure apparatus

Despite significant accuracy improvement in convolutional neural networks (CNN) based object detectors, they often require prohibitive runtimes to process an image for real-time applications. State-of-the-art models often use very deep networks with a large number of floating point operations. Efforts such as model compression learn compact models with fewer number of parameters, but with much reduced accuracy. In this work, we propose a new framework to learn compact and fast object detection networks with improved accuracy using knowledge distillation [20] and hint learning [34]. Although knowledge distillation has demonstrated excellent improvements for simpler classification setups, the complexity of detection poses new challenges in the form of regression, region proposals and less voluminous labels. We address this through several innovations such as a weighted cross-entropy loss to address class imbalance, a teacher bounded loss to handle the regression component and adaptation layers to better learn from intermediate teacher distributions. We conduct comprehensive empirical evaluation with different distillation configurations over multiple datasets including PASCAL, KITTI, ILSVRC and MS-COCO. Our results show consistent improvement in accuracy-speed trade-offs for modern multi-class detection models.

/pdf/learning-efficient-object-detection-models-with-knowledge-123cmg4byv.pdf

Learning efficient object detection models with knowledge distillation

Suggested by the structure of the visual nervous system, a new algorithm is proposed for pattern recognition. This algorithm can be realized with a multilayered network consisting of neuron-like cells. The network, “neocognitron”, is self-organized by unsupervised learning, and acquires the ability to recognize stimulus patterns according to the differences in their shapes: Any patterns which we human beings judge to be alike are also judged to be of the same category by the neocognitron. The neocognitron recognizes stimulus patterns correctly without being affected by shifts in position or even by considerable distortions in shape of the stimulus patterns.

Neocognitron--A New Algorithm for Pattern Recognition Tolerant of Deformations and Shifts in Position

This invention provides an object extraction method for performing processing for extracting and cutting out a specific object from a sensed image at high speed, and an image sensing apparatus using the method. In this invention, in a method of extracting an object by comparing a sensed image and a standard image, a focusing signal, focal length data, visual axis direction data, and illumination conditions are detected, and the initial size, initial position, or initial color of the standard image is changed on the basis of the detection results, and extraction is started under optimal conditions. In a method of extracting a specific object from the background image, the background image is converted into an image having the same conditions as those of the object image. From a plurality of images obtained under different image sensing conditions, the contour of the object is accurately obtained at high speed. In an object extraction method using a template, the size of the template is determined on the basis of the object distance, object size, or the like.

Object extraction method, and image sensing apparatus using the method

This invention relates to an image processing method for precisely and automatically recognizing a specific image extraction region to be divided from an input image, and extracting the specific image extraction region. Thresholding is done for intensity differences of edge data obtained from the input data to extract difference edge data. A main extraction region is estimated from the outermost contour line extracted based on the extraction result of the difference edge data. Thresholding or the like is done in units of pixels of the input image to extract an initial region. Extraction region determination processing is done for extracting an accurate target region by combining the main extraction region and initial region.

Method of extracting image from input image using reference image

An image recognition/reproduction method includes an extraction step of extracting local feature elements of an image, and a selection step of selecting a pair composed of a prescribed local feature element and position information indicative thereof, this pair being such that the distance between a pair composed of a prescribed local feature element and position information indicative thereof and a pair composed of a local feature element extracted at the extraction step and position information indicative thereof is less than a prescribed distance.

Image recognition/reproduction method and apparatus

In a hierarchical neural network having a module structure, learning necessary for detection of a new feature class is executed by a processing module which has not finished learning yet and includes a plurality of neurons which should learn an unlearned feature class and have an undetermined receptor field structure by presenting a predetermined pattern to a data input layer. Thus, a feature class necessary for subject recognition can be learned automatically and efficiently.

Information processing apparatus, information processing method, pattern recognition apparatus, and pattern recognition method

Images sensed through object lenses 100 R and 100 L, having zoom lenses, with image sensors 102 R and 102 L are processed by image signal processors 104 R and 104 L, and an image of an object in each of the sensed images is separated from a background image on the basis of the processed image signals. The separated image signals representing the image of the object enter the image processor 220 , where a three-dimensional shape of the object is extracted on the basis of parameters used upon sensing the images. The parameters are automatically adjusted so that images of the object fall within the both image sensing areas of the image sensors 102 R and 102 L and that they fall within the both focal depths of the image sensors 102 R and 102 L.

Masakazu Matsugu

Papers

Object extraction method, and image sensing apparatus using the method

Method of extracting image from input image using reference image

Image recognition/reproduction method and apparatus

Information processing apparatus, information processing method, pattern recognition apparatus, and pattern recognition method

Image sensing and image processing apparatuses