公开(公告)号：US20220004855A1

公开(公告)日：2022-01-06

申请号：US17290519

申请日：2019-11-04

Applicant: BEIJING HORIZON ROBOTICS TECHNOLOGY RESEARCH AND DEVELOPMENT CO., LTD.

Inventor： Haoqian HE ,  Jianjun LI ,  Chang HUANG

IPC: G06N3/063 ,  G06N3/04

Abstract: Disclosed are a convolution processing engine and a control method thereof, and a convolutional neural network accelerator comprising the convolution processing engine. The convolution processing engine comprises at least two cache memories connected in series and an operational circuit. The convolution processing engine can realize an efficient convolution operation with lower complexity and power consumption.

公开(公告)号：US20230409886A1

公开(公告)日：2023-12-21

申请号：US18247408

申请日：2022-02-10

Applicant: Beijing Horizon Robotics Technology Research and Development Co., Ltd.

Inventor： Zhuoran ZHAO ,  Kai YU ,  Chang HUANG ,  Zhenjiang WANG ,  Jianjun LI ,  Delin LI ,  Yinan ZHANG

IPC: G06N3/0464

CPC classification number: G06N3/0464

Abstract: The present disclosure provides a method and apparatus for deconvolving feature data using convolution hardware. The method includes: reading a feature map and deconvolution kernel into on-chip memory, and padding zeroes to the feature map; determining convolution kernels based on the deconvolution kernel; removing a row and/or column of each convolution kernel whose elements all are invalid weights, to obtain an optimized convolution kernel, and removing a corresponding row and/or column in the zero-padded feature map to obtain an corresponding optimized feature map; convolving each optimized convolution kernel with corresponding optimized feature map using the multiply-add array, to obtain convolutional outputs; and interleaving and synthesizing the convolutional outputs to obtain an interleaving synthetic output including at least a deconvolutional output corresponding to the feature map and deconvolution kernel. The method reduces hardware complexity, chip area and power consumption, and many invalid operations, improving operating efficiency of convolution hardware.

公开(公告)号：US20200285911A1

公开(公告)日：2020-09-10

申请号：US16788661

申请日：2020-02-12

Applicant: Beijing Horizon Robotics Technology Research and Development Co., Ltd.

Inventor： Chaoxu GUO ,  Qian ZHANG ,  Guoli WANG ,  Chang HUANG

IPC: G06K9/62 ,  G06N3/04

Abstract: An image recognition method includes: determining a first feature map of the current frame image by using a convolutional neural network based on a type of a current frame image; determining a second feature map of a key frame image before the current frame image; performing feature alignment on the first feature map and the second feature map to obtain a first aligned feature map; fusing the first feature map and the first aligned feature map to obtain a first fused feature map; and recognizing content in the current frame image based on the first fused feature map.