Abstract: A semiconductor package apparatus may include technology to apply a trained scene text detection network to an image to identify a core text region, a supportive text region, and a background region of the image, and detect text in the image based on the identified core text region and supportive text region. Other embodiments are disclosed and claimed.

Abstract: Autonomous robots and methods of operating the same are disclosed. An autonomous robot includes a sensor and memory including machine readable instructions. The autonomous robot further includes at least one processor to execute the instructions to generate a velocity costmap associated with an environment in which the robot is located. The processor generates the velocity costmap based on a source image captured by the sensor. The velocity costmap includes velocity information indicative of movement of an obstacle detected in the environment.

Abstract: An installation assembly of a ceiling lighting fixture, a ceiling lighting fixture and a ceiling lamp are provided. The installation assembly includes a fixing base and a connecting base detachably connected with the fixing base. One of the fixing base and the connecting base is provided with a snap recess, and the other of the fixing base and the connecting base is provided with an elastic buckle which is capable of being snap-fitted with the snap recess. The elastic buckle has a guide slope, the elastic buckle is capable of being compressed as the connecting base moves toward the fixing base, and being introduced into the snap recess through the guide slope. The elastic buckle is rotationally fitted with the snap recess, and the snap recess is provided with an avoidance notch for the elastic buckle to be detached.

Abstract: A mechanism is described for facilitating slimming of neural networks in machine learning environments. A method of embodiments, as described herein, includes learning a first neural network associated with machine learning processes to be performed by a processor of a computing device, where learning includes analyzing a plurality of channels associated with one or more layers of the first neural network. The method may further include computing a plurality of scaling factors to be associated with the plurality of channels such that each channel is assigned a scaling factor, wherein each scaling factor to indicate relevance of a corresponding channel within the first neural network. The method may further include pruning the first neural network into a second neural network by removing one or more channels of the plurality of channels having low relevance as indicated by one or more scaling factors of the plurality of scaling factors assigned to the one or more channels.

Abstract: Video analysis may be used to determine who is watching television and their level of interest in the current programming Lists of favorite programs may be derived for each of a plurality of viewers of programming on the same television receiver.

Abstract: A mechanism is described for facilitating slimming of neural networks in machine learning environments. A method of embodiments, as described herein, includes learning a first neural network associated with machine learning processes to be performed by a processor of a computing device, where learning includes analyzing a plurality of channels associated with one or more layers of the first neural network. The method may further include computing a plurality of scaling factors to be associated with the plurality of channels such that each channel is assigned a scaling factor, wherein each scaling factor to indicate relevance of a corresponding channel within the first neural network. The method may further include pruning the first neural network into a second neural network by removing one or more channels of the plurality of channels having low relevance as indicated by one or more scaling factors of the plurality of scaling factors assigned to the one or more channels.

Abstract: A passenger conveying device, including: an entrance, an exit, and a plurality of steps disposed between the entrance and the exit. A skirt guard panel is respectively disposed on two sides of the step, and the step is constructed to reciprocally move between the entrance and the exit along a path defined by the skirt guard panel. The step includes a tread, a tread guard panel is attached to each of two sides of tread, and the tread guard panel is located between the tread and the skirt guard panel. The tread guard panel includes two or more protrusions, each protrusion is constructed to protrude with respect to an upper surface of the tread, and a protrusion farther away from the tread has a larger protrusion height. The passenger conveying device of the present invention has the advantages of simple structure, reliable installation, and convenient maintenance.

Abstract: Systems, apparatuses and methods may provide for detecting a facial image including generating a spatial convolutional neural network score for one or more detected facial images from a facial image detector, generating a temporal convolutional network score for detected facial video frames from the facial image detector and generating a combined spatial-temporal score to determine whether a detected facial image gains user access to a protected resource.

Abstract: Techniques related to implementing convolutional neural networks for object recognition are discussed. Such techniques may include generating a set of binary neural features via convolutional neural network layers based on input image data and applying a strong classifier to the set of binary neural features to generate an object label for the input image data.

Abstract: One embodiment provides for a compute apparatus to perform machine learning operations, the compute apparatus comprising instruction decode logic to decode a single instruction including multiple operands into a single decoded instruction, the multiple operands including a first operand and a second operand, the first operand including vector of one-hot coded weights and the second operand including a vector of input data; and a general-purpose graphics compute unit including a first logic unit, the general-purpose graphics compute unit to execute the single decoded instruction, wherein to execute the single decoded instruction includes to perform multiple operations on the first set of operands and the second set of operands.

Abstract: An electrical connector and luminaire assembly are provided. The electrical connector includes a base including a bottom wall and two side walls, the two side walls being on both sides of the bottom wall, and the bottom wall and the two side walls defining a receiving channel; a first printed circuit board fixed on the base, the first printed circuit board and the receiving channel being respectively located at two sides of the bottom wall; and a plurality of first conductive terminals electrically connected to the first printed circuit board and extending through the bottom wall of the base into the receiving channel.

Abstract: An automatic handrail tensioning system and a method for adjusting the tension degree of a handrail, and belongs to the technical field of escalators. The automatic handrail tensioning system of the present invention comprises: a sensor for detecting information that can reflect a tension degree of the handrail; a controller for determining the tension degree information of the handrail according to the information detected by the sensor, and generating a corresponding control instruction for adjusting the tension degree of the handrail based on the tension degree information; and an actuator for driving a tensioning device to adjust the tension degree of the handrail based on the control instruction.

Abstract: The present disclosure provides a remote controller, a lighting device and a lighting system. The remote controller is configured for controlling color of light emitted by a target lighting device and includes: a housing; a main control panel of a remote control terminal, located in the housing and provided with a color acquiring circuit for acquiring color information of a target region, and a wireless transmitter for transmitting the color information to the target lighting device; a color acquiring switch, located on the housing and electrically connected with both the color acquiring circuit and the wireless transmitter, and the color acquiring switch being configured for turning on/off the color acquiring circuit and the wireless transmitter. The remote controller has a simple structure and convenient operation, so that the color of the light emitted by the lighting device may be freely adjusted according to individual needs, which is convenient.

Abstract: Provided a complex of formula is [3-((1S, 3R)-1-biphenyl-4-ylmethyl-3-ethoxycarbonyl-1-butylcarbamoyl) propionate-(S)-3?-methyl-2?-(pentanoyl {2?-(tetrazol-5-ylate) biphenyl-4?-ylmethyl}amino) butyrate]6.XCa2+.YNa+.ZH2O, wherein X=1-3, Y=12-16, Z=9-18, and 2X+Y=18, and represented by formula (I). Also disclosed are the method of preparing the complex and the method of treating chronic heart disease using a medicament comprising the complex.

Abstract: An example apparatus for semantic image segmentation includes a receiver to receive an image to be segmented. The apparatus also includes a gated dense pyramid network comprising a plurality of gated dense pyramid (GDP) blocks to be trained to generate semantic labels for each pixel in the received image. The apparatus further includes a generator to generate a segmented image based on the generated semantic labels.

Abstract: Video analysis may be used to determine who is watching television and their level of interest in the current programming Lists of favorite programs may be derived for each of a plurality of viewers of programming on the same television receiver.

Abstract: An artificial hip joint with a composite ball head includes an artificial acetabulum (1) and a composite ball head (2); wherein an internal surface of the artificial acetabulum (1) cooperates with an external surface of the composite ball head (2), and the composite ball head (2) can rotate within the artificial acetabulum (1); the internal surface of the artificial acetabulum (1) is directly in contact with the composite ball head (2) without a liner therebetween; the composite ball head (2) comprises a metal ball head (22) and a non-metallic shell (21) wrapped around an external surface of the metal ball head (22). For the composite ball head (2), the metal ball head (22) is wrapped with a non-metallic shell (21), which has large supporting force capability, has strong bearing capacity, effectively reduces the contact stress of the non-metallic shell (21), is stable and reliable.

Abstract: A color picking device and a color picking remote controller are provided. The color picking device is configured to acquire a color of a target object, which includes: a substrate; a light emitting module configured to emit white light toward the target object, where the light emitting module that includes a light source and a light distribution element; and a color receiving module configured to receive light reflected by the target object and having the color of the target object after the target object is illuminated by the white light, where the color receiving module includes a first housing, a lens assembly disposed in the first housing, and the sensing element, where the lens assembly includes a first convex lens, an aperture, a concave lens and a second convex lens that are coaxially arranged from top to bottom along a top of the first housing.

Abstract: A lens is provided. The lens includes a bottom surface, an inner surface, an outer surface, and a first receiving chamber configured to receive a light-emitting component; the inner surface includes one first light-incident surface and two second light-incident surfaces, the two second light-incident surfaces are opposite to each other, the first light-incident surface is a curved surface or a planar surface, the bottom surface is a surface adjacent to the second light-incident, the outer surface includes two first reflective surfaces forming a V-shaped structure; two first light-exiting surfaces located at both sides of the two first reflective surfaces; and two second light-exiting surfaces adjacent to the first light-exiting surfaces.

Abstract: Techniques are provided for training and operation of a topic-guided image captioning system. A methodology implementing the techniques according to an embodiment includes generating image feature vectors, for an image to be captioned, based on application of a convolutional neural network (CNN) to the image. The method further includes generating the caption based on application of a recurrent neural network (RNN) to the image feature vectors. The RNN is configured as a long short-term memory (LSTM) RNN. The method further includes training the LSTM RNN with training images and associated training captions. The training is based on a combination of: feature vectors of the training image; feature vectors of the associated training caption; and a multimodal compact bilinear (MCB) pooling of the training caption feature vectors and an estimated topic of the training image. The estimated topic is generated by an application of the CNN to the training image.