Huawei Technologies Co., Limited

Zhejiang University

Yingchun Yang

Jianwei Yin

Shuiguang Deng

Liang Song

Yaoda Liu

Yourui Huangfu

Li Chen

Wenxiong Wei

Zhaohui Wu

Ying Li

Yongheng Shang

Chen Qiu

Ge Su

Shuncheng Pan

Shaoqing Liu

Yuxiang Cai

The present invention discloses a semantic segmentation method for cross-satellite remote sensing images based on unsupervised bidirectional domain adaptation and fusion. The method includes training of bidirectional source-target domain image translation models, selection of bidirectional generators in the image translation models, bidirectional translation of source-target domain images, training of source and target domain semantic segmentation models, and generation and fusion of source and target domain segmentation probabilities. According to the present invention, by utilizing source-target and target-source bidirectional domain adaptation, the source and target domain segmentation probabilities are fused, which improves the accuracy and robustness of a semantic segmentation model for the cross-satellite remote sensing images; and further, through the bidirectional semantic consistency loss and the selection of the parameters of the generators, the influence due to the instability problem of the generators in the bidirectional image translation models is avoided.

Semantic segmentation method for cross-satellite remote sensing images based on unsupervised bidirectional domain adaptation and fusion

The present disclosure discloses an unsupervised domain adaptation method, a device, a system and a storage medium of semantic segmentation based on uniform clustering; first, a prototype-based source domain uniform clustering loss and an empirical prototype-based target domain uniform clustering loss are established, to reduce intra-class differences of pixels responding to the same category; meanwhile, the pixels with similar structures but different classes are driven away from each other, wherein they tend to be evenly distributed, increasing the inter-class distance and overcoming the problem that the category boundaries are unclear during the domain adaptation process; next, the prototype-based source domain uniform clustering loss and the empirical prototype-based target domain uniform clustering loss are integrated into an adversarial training framework, which reduces the domain difference between the source domain and the target domain, thus improving the accuracy of semantic segmentation.

Unsupervised domain adaptation method, device, system and storage medium of semantic segmentation based on uniform clustering

An on-chip optical interconnection structure and network, where the on-chip optical interconnection structure includes M levels of optical switches, and m-level optical switches in the M levels of optical switching devices include 2optical switches. Each optical switch in (i-1)-level optical switches in the M levels of optical switches is coupled to two optical switches in i-level optical switches. Two optical switches in the i-level optical switches coupled to a same optical switch in the (i-1)-level optical switches are coupled. The on-chip optical interconnection network is divided into levels, and switches coupled in a grid manner are formed such that hierarchical switching may be performed, and conflicts and delays in communication are reduced.

On-chip optical interconnection structure and network

A signal transmission method includes receiving, by a controller, a request signal sent by a first transmitter. The method includes establishing, by the controller according to the request signal, a second optical path that connects the first transmitter to the second receiver in the optical switching network (OSN). The method includes sending, by the controller, to the second receiver according to the request signal, a reset signal used to instruct the second receiver to be reset to an initial state. The method includes sending, by the controller, an acknowledgement signal to the first transmitter. The acknowledgment signal is used to instruct the first transmitter to send, by using the second optical path, an optical signal to the second receiver.

Signal transmission method, controller, and signal transmission system

An optical buffer and a method for storing an optical signal using the optical buffer, where the optical buffer includes a first waveguide, a first optical delay waveguide loop and a controller. The first waveguide includes a first arm and a second arm, where a first end of the first arm is an input end of the optical buffer, and a second end of the second arm is an output end of the optical buffer. A second end of the first arm connects to a first end of the second arm. The first optical delay waveguide loop connects to the first arm at a first end using a first optical switch, and a second part of the first optical delay waveguide loop connects to the second arm at a second end using a second optical switch. The controller connects to the first optical switch and the second optical switch respectively.

Optical buffer and methods for storing optical signal

An annular optical shifter and a method for controlling shift, where the annular optical shifter includes: a first bent straight-through waveguide, connecting an input end and an output end of an optical signal, and configured to transmit, to the output end, the optical signal input from the input end; multiple optical delay waveguide loops, arranged transversely and parallel on two arms of the first bent straight-through waveguide, where the multiple optical delay waveguide loops are configured to temporarily store optical signals; multiple pairs of optical switches, where each pair of optical switches are configured to control on and off of an optical path that is on the two arms of the first bent straight-through waveguide and two sides of an optical delay waveguide loop corresponding to each pair of optical switches; and a controller, configured to implement shift-up or shift-down of the optical signals.

Annular optical shifter and method for shifting optical signal

A speaker recognition method through emotional model synthesis based on Neighbors Preserving Principle is enclosed. The methods includes the following steps: (1) training the reference speaker's and user's speech models; (2) extracting the neutral-to-emotion transformation/mapping sets of GMM reference models; (3) extracting the emotion reference Gaussian components mapped by or corresponding to several Gaussian neutral reference Gaussian components close to the user's neutral training Gaussian component; (4) synthesizing the user's emotion training Gaussian component and then synthesizing the user's emotion training model; (5) synthesizing all user's GMM training models; (6) inputting test speech and conducting the identification. This invention extracts several reference speeches similar to the neutral training speech of a user from a speech library by employing neighbor preserving principles based on KL divergence and combines an emotion training speech of the user using the emotion reference speech in the reference speech, improving the performance of the speaker recognition system in the situation where the training speech and the test speech are mismatched, and the robustness of the speaker recognition system is increased.

Growing community of inventors

Hangzhou, China

Yingchun Yang