Abstract: A convolutional neural network framework is described that uses reverse connection and obviousness priors for object detection. A method includes performing a plurality of layers of convolutions and reverse connections on a received image to generate a plurality of feature maps, determining an objectness confidence for candidate bounding boxes based on outputs of an objectness prior, determining a joint loss function for each candidate bounding box by combining an objectness loss, a bounding box regression loss and a classification loss, calculating network gradients over positive boxes and negative boxes, updating network parameters within candidate bounding boxes using the joint loss function, repeating performing the convolutions through to updating network parameters until the training converges, and outputting network parameters for object detection based on the training images.

Abstract: A lithium-based energy storage system includes an electrolyte and an electrode. The electrode has a conformal coating of parylene. The parylene forms an artificial solid electrolyte interface (SEI). The electrode may include a material chosen from silicon, graphene-silicon composite, carbon-sulfur, and lithium. The use of parylene to form a conformal coating on an electrode in a lithium-based energy storage system is also disclosed.

Abstract: The present invention relates to a method for activating tumor-infiltrating lymphocytes (TILs) in a subject in need, particularly by administering to the subject a fermented composition generated via fermentation of symbiotic microbiota in a medium.

Abstract: An object tracking method includes: obtaining at least one image acquired by at least one image acquisition device; obtaining a first appearance feature of a target object and a first spatial-temporal feature of the target object based on the at least one image; obtaining an appearance similarity and a spatial-temporal similarity between the target object and each global tracking object in a currently recorded global tracking object queue; based on determining that the target object matches a target global tracking object based on the appearance similarity and the spatial-temporal similarity, allocating a target global identifier corresponding to the target global tracking object to the target object; determining, using the target global identifier, a plurality of associated images acquired by a plurality of image acquisition devices associated with the target object; and generating, based on the plurality of associated images, a tracking trajectory matching the target object.

Abstract: A method for accessing a data bus includes setting a first-come-first-served basis for determining priorities between masters in addition to a fixed priority being set between the same masters. A number of master ports are connected to a number of masters, and a number of slave ports are connected to a number of slaves. First and second multiplexers are connected between the master ports and the slave ports, a number of decoders are connected to the second multiplexers, and a number of arbiters are connected to the first multiplexers. The master ports have a fixed priority, but each arbiter, in receiving an access-request signal sent by a master port, can determine an order as to which of multiple master ports can access a slave port according to a combination of the fixed priority basis and the first-come-first-served basis. A system and a relevant device are also disclosed.

Abstract: A gantry-less particle therapy system is provided. Charged particles are extracted from an ion source and accelerated in a beam transport system having an annular portion extending in a first plane and that circumscribes a volume, an arcuate portion extending in a second plane, and a transition portion that connects the annular portion and the arcuate portion. The arcuate portion terminates at a beam nozzle extending radially inward from the annular portion to deliver an ion beam to a treatment area contained in the volume circumscribed by the annular portion.

Abstract: The present disclosure provides an ultrasonic drive and driving method configured for driving an ultrasonic tool. The ultrasonic drive includes a switch module, a sensing element and a control element. The sensing element senses the voltage and current of the ultrasonic tool and generates a sensing signal accordingly. The control element receives the sensing signal and outputs a control signal. The switch module outputs an ultrasonic signal according to the control signal for controlling the vibration of the ultrasonic tool. When the ultrasonic drive operates a frequency sweep function, the control element determines an operating interval and an operating frequency of the ultrasonic signal. When the ultrasonic drive operates a frequency following function, the control element adjusts the operating frequency according to the sensing signal for keeping the impedance of the ultrasonic tool consistent.

Abstract: A multi-camera architecture for detecting and tracking a ball in real-time. The multi-camera architecture includes network interface circuitry to receive a plurality of real-time videos taken from a plurality of high-resolution cameras. Each of the high-resolution cameras simultaneously captures a sports event, wherein each of the plurality of high-resolution cameras includes a viewpoint that covers an entire playing field where the sports event is played. The multi-camera architecture further includes one or more processors coupled to the network interface circuitry and one or more memory devices coupled to the one or more processors.

Abstract: An image classification system includes a storage device, a computing device and a first processing device. The storage device stores a plurality of pseudo-centroid datasets, wherein the pseudo-centroid datasets correspond to a plurality of units of first image dataset, and the number of pseudo-centroid data points of each of the pseudo-centroid datasets is much smaller than the number of data points of each of the units of first image dataset. The computing device receives the second image data and computes a plurality of feature values of the second image data. The first processing device receives the feature values and the pseudo-centroid datasets, and compares the feature values with the pseudo-centroid data points to identify and classify the second image data.

Abstract: A head mounted display system includes a scanning unit and a processing unit. The scanning unit is configured to scan a real object in a real environment so as to generate a scanning result. The processing unit is coupled to the scanning unit. The processing unit is configured to identify the real object according to the scanning result of the scanning unit, determine at least one predetermined interactive characteristic according to an identification result of the processing unit, create a virtual object in a virtual environment corresponding to the real object in the real environment according to the scanning result of the scanning unit, and assign the at least one predetermined interactive characteristic to the virtual object in the virtual environment. Therefore, the present disclosure allows a user to manipulate the virtual object in different ways more naturally, which effectively improves the user's interactive experience.

Abstract: A cordless roller shutter structure includes a roller shutter fixing seat, a sleeve controller, a tension adjuster and a shutter sheet locating unit. The sleeve controller is fitted in a sleeve and securely disposed on the roller shutter fixing seat. The shutter sheet is controlled to suspend downward or roll upward directly by means of pulling and releasing the lower weight of the shutter sheet without using the bead chain. A bearing is received in each of two collars so that the sleeve can more easily drive the shutter sheet to save strength. In addition, by means of the adjustment disc of the tension adjuster and the shutter sheet locating unit, the ascending/descending speed of the shutter sheet can be adjusted and the sunshade position of the shutter sheet can be locked. Accordingly, the utility and effect of the cordless roller shutter are enhanced.

Abstract: An electronic device includes a motor, a sensing device and a controller. The motor provides a motive power to the load. The sensing device measures a plurality of motion parameters of the motor. The controller calculates a real system parameter between the motor and the load according to the motion command and the plurality of motion parameters. According to the motion command and the plurality of motion parameters, the controller acquires a real value of a rotation speed of the motor. Moreover, controller establishes a system model according to the motion command and the real system parameter so as to acquire an estimated value of the rotation speed of the motor. If the first mean square error is not smaller than the first target value, the controller performs a particle swarm optimization algorithm to change the estimated value of the rotation speed of the motor.

Abstract: A production system, a drive and an operating method provide the functions of quality measurement and mechanism diagnosis. The production system has a hierarchical processing structure. The drive includes a real system driving module and a virtual system driving module. The real system driving module generates a mechanism real operation parameter information. In a system identification mode, the drive creates at least one virtual mechanism model. The drive generates a mechanism stimulation operation parameter information according to a processing strategy of a controller, the mechanism real operation parameter information and the at least one virtual mechanism model. The controller performs a quality measurement operation, performs a mechanism diagnosis operation and/or adjusts the processing strategy according to the mechanism real operation parameter information and the mechanism stimulation operation parameter information.

Abstract: An adaptive learning power modeling method includes: sampling at least one of a plurality of network components to form a power consumption evaluation network according to at least one parameter within a parameter range; evaluating a predictive power consumption of a to-be-measured circuit by the power consumption evaluation network; training and evaluating an actual power consumption and the predictive power consumption of the to-be-measured circuit by the power consumption evaluation network to obtain an evaluation result; and performing training according to the evaluation result to determine whether to change the power consumption evaluation network.

Abstract: A deep neural network (DNN) hardware accelerator including a processing element array is disclosed. The processing element array includes a processing element array, the processing element array including a plurality of processing element groups and each of the processing element groups including a plurality of processing elements. A first network connection implementation between a first processing element group of the processing element groups and a second processing element group of the processing element groups is different from a second network connection implementation between the processing elements in the first processing element group.

Abstract: An image classification system includes a storage device, a computing device and a first processing device. The storage device stores a plurality of pseudo-centroid datasets, wherein the pseudo-centroid datasets correspond to a plurality of units of first image dataset, and the number of pseudo-centroid data points of each of the pseudo-centroid datasets is much smaller than the number of data points of each of the units of first image dataset. The computing device receives the second image data and computes a plurality of feature values of the second image data. The first processing device receives the feature values and the pseudo-centroid datasets, and compares the feature values with the pseudo-centroid data points to identify and classify the second image data.