Nextvpu (Shanghai) Co., Ltd.

Semiconductor Manufacturing International (Shanghai) Corporation

Xinpeng Feng

Ji Zhou

Haijiao Cai

Shu Fang

Chang Sun

Yin Tang

Zhijun Lyu

Peng Zhang

Di Peng

Henry Shi Li

Jun Pang

Song Mei

Xuanyang Feng

Gang Qin

Linfeng Li

Aofeng Qian

Zhijun Lv

Data processing reader

Provided are a method for pre-allocating memory, an electronic device, and a medium. The method comprises: attempting to obtain a memory block identifier from a list of idle memory block identifiers for each output from each layer; if the memory block identifier is obtained, storing an identifier of the output and the memory block identifier into a mapping table and deleting the memory block identifier from the list of idle memory block identifiers; or if not obtained, creating a new memory block identifier, storing an identifier of the output and the new memory block identifier into a mapping table, and putting the new memory block identifier into a general list of memory block identifiers; for each input to a current layer, putting a memory block identifier corresponding to the input into the list of idle memory block identifiers if the input is not used as an input to a subsequent layer.

Memory pre-allocation for forward calculation in a neural network

A wiring design method and a wiring structure for a package substrate in a flip chip, and a flip chip. The wiring design method includes: arranging bump pads in an array of rows and columns, wherein the bump pads are configured to bond with conductive bumps on a flip chip die, and the bump pads comprise signal pads and non-signal pads; providing the non-signal pad with a via hole; and using a layer of wiring to lead a subset of the signal pads out of an orthographic projection region of the flip chip die on the package substrate, wherein the subset of the signal pads is configured to carry all functional signals required by design specifications of the flip chip die for the array of the bump pads.

Wiring design method, wiring structure, and flip chip

An image text broadcasting method, an electronic device, and a storage medium are provided. The image text broadcasting method includes: receiving a specified broadcast indication; determining a current broadcast progress about broadcast data in response to the specified broadcast indication; acquiring a next piece of broadcast data from a first text according to the current broadcast progress and the specified broadcast indication, in which the first text is composed of text data recognized and stored for a text in a text area of an image.

Image text broadcasting

A layout analysis method, an electronic device, and a non-transitory computer-readable storage medium are provided. The layout analysis method includes: obtaining coordinate information of a plurality of text lines in an image; creating a layout model of the image according to the coordinate information; analyzing a layout structure of the text lines based on the layout model; and determining an order of the text lines relative to each other based on the layout structure.

Layout analysis

Disclosed is a compression method for a deep neural network. The deep neural network comprises a plurality of layers. The method includes the following steps for each of at least one layer of the plurality of layers other than an input layer: reading parameters of that layer from a parameter storage space; reading a first feature map stored by an immediate previous layer from a feature map storage space; generating a second feature map on the basis of the read parameters and the first feature map; compressing the second feature map so as to obtain a third feature map; and storing the third feature map into the feature map storage space. Further, a chip based on the compression method and a medium are also provided.

Compression for deep neural network

An image depth determining method and a living body identification method, a circuit, a device, and a medium. The image depth determining method includes obtaining pixel coordinates of a feature point pair associated with an object point on a subject, determining a first straight line passing through the origin of a first camera coordinate system of a first camera based on first pixel coordinates and intrinsic parameters of the first camera, determining a second straight line passing through the origin of a second camera coordinate system of a second camera based on second pixel coordinates and intrinsic parameters of the second camera, and determining the depth of the object point based on the first straight line, the second straight line, and extrinsic parameters describing a relative position relationship between the first camera and the second camera.

Image depth determining method and living body identification method, circuit, device, and medium

A disparity estimation method, an electronic device, and a computer-readable storage medium are provided. The disparity estimation method includes: performing feature extraction on each image in an image pair; and performing cascaded multi-stage disparity processing according to the extracted image features to obtain multiple disparity maps with increasing sizes. The input of a first stage disparity processing in the multi-stage disparity processing includes multiple image features each having a size corresponding to the first stage disparity processing, and the input of disparity processing of each stage other than the first stage disparity processing in the multi-stage disparity processing includes: one or more image features each having a size corresponding to disparity processing of the stage and a disparity map generated by disparity processing of an immediate previous stage.

Disparity estimation

A disparity determining method, an electronic device, and a computer-readable storage medium are provided. The disparity determining method includes: downsampling a first image and a second image, to obtain a first image with a first size and a second image with the first size; performing first-stage disparity computation on the first image with the first size and the second image with the first size, to obtain a first disparity map with the first size; and performing second-stage disparity computation on a first image with a second size and a second image with the second size based on an upsampled disparity map, with the second size, of the first disparity map, to obtain a second disparity map with the second size.

Determination of disparity

The present disclosure relates to the field of wearable devices, and in particular to a connector for a wearable device, an assistive device, a wearable device, and a kit. The connector comprises a first connecting component and a second connecting component, wherein the first connecting component comprises a first engaging member, the second connecting component comprises a second engaging member, and the first connecting component and the second connecting component are connected to each other by means of the mating between the first engaging member and the second engaging member and can rotate relative to each other into a separable position and an engaged position. The connector of the present disclosure can solve the technical problem that the mounting angle of the assistive device relative to the wearable device cannot be flexibly and conveniently adjusted and cannot be maintained in each adjustment shift.

Adjustable connection between an assistive device and a wearable device

A mobile robot is provided. The mobile robot includes: a driving mechanism configured to drive the mobile robot to move; a chassis on which the driving mechanism is mounted; and a camera mainboard, on which a camera module is provided. The camera module is configured to obtain an external environment image and has at least two optical axes. The camera mainboard is arranged on the chassis such that each of the at least two optical axes is parallel to a horizontal plane.

Mobile robot

Electronic device for finding and identifying objects

The present disclosure relates to a connector, an assistive device, a wearable device, and a wearable device kit. The connector includes a first magnetic connecting component; and a second magnetic connecting component detachably connected to the first magnetic connecting component by magnetic attraction. By means of the connector of the present disclosure, the assistive device can be detachably connected to the wearable device more reliably and more flexibly.

Growing community of inventors

Shanghai, China

Xinpeng Feng