-
公开(公告)号：US11436435B2
公开(公告)日：2022-09-06
申请号：US16985170
申请日：2020-08-04
Inventor： Anping Li , Shaoxin Li , Chao Chen , Pengcheng Shen , Shuang Wu , Jilin Li
Abstract: This application relates to a model training method. The method includes retrieving a current group of training samples, the training samples being based on a training set; obtaining first sample features of training samples in the current group of training samples based on a to-be-trained model; and obtaining, center features respectively corresponding to the training samples; obtaining feature distribution parameters corresponding to the training samples, the feature distribution parameter corresponding to each training sample being obtained by collecting statistics on second sample features of training samples in the training set that belong to the same classification category, and the second sample feature of each training sample being generated by a trained model; obtaining, based on the center features and the feature distribution parameters, a comprehensive loss parameter corresponding to the current group of training samples; and adjusting model parameters of the to-be-trained model based on the comprehensive loss parameter.
-
Patent Agency Ranking
公开(公告)号：US11335124B2
公开(公告)日：2022-05-17
申请号：US16927812
申请日：2020-07-13
Inventor： Anping Li , Shaoxin Li , Chao Chen , Pengcheng Shen , Jilin Li
Abstract: This application relates to a face recognition method performed at a computer server. After obtaining a to-be-recognized face image, the server inputs the to-be-recognized face image into a classification model. The server then obtains a recognition result of the to-be-recognized face image through the classification model. The classification model is obtained by inputting a training sample marked with class information into the classification model, outputting an output result of the training sample, calculating a loss of the classification model in a training process according to the output result, the class information and model parameters of the classification model, and performing back propagation optimization on the classification model according to the loss.
-
3.发明申请公开(公告)号：US20200342214A1
公开(公告)日：2020-10-29
申请号：US16927812
申请日：2020-07-13
Inventor： Anping LI , Shaoxin Li , Chao Chen , Pengchen Shen , Jilin Li
Abstract: This application relates to a face recognition method performed at a computer server. After obtaining a to-be-recognized face image, the server inputs the to-be-recognized face image into a classification model. The server then obtains a recognition result of the to-be-recognized face image through the classification model. The classification model is obtained by inputting a training sample marked with class information into the classification model, outputting an output result of the training sample, calculating a loss of the classification model in a training process according to the output result, the class information and model parameters of the classification model, and performing back propagation optimization on the classification model according to the loss.
-
公开(公告)号：US11210810B2
公开(公告)日：2021-12-28
申请号：US16928427
申请日：2020-07-14
Abstract: A camera localization method includes: obtaining an environment map of a target environment, predicting a location of a camera when shooting a target image according to location information of the camera when shooting a history image before the target image is shot to obtain predicted location information of the camera; filtering out at least one feature point that is currently not observable by the camera in the environment map according to the predicted location information of the camera, location information of each feature point and viewing-angle area information of each feature point in the environment map; and matching the feature point in the target image with remaining feature points in the environment map after the filtering to obtain a feature point correspondence, and determining location information of the camera according to the feature point correspondence.
-
公开(公告)号：US11142121B2
公开(公告)日：2021-10-12
申请号：US16597484
申请日：2019-10-09
Inventor： Chao Chen , Wei Wu , Chengjun Li
IPC: B60Q1/50 , G05B19/042 , G05D1/02
Abstract: Aspects of the disclosure provide methods and mobile robots for providing information to a pedestrian. In some examples, a mobile robot includes processing circuitry. The processing circuitry obtains, via a sensor, sensing data that indicates an ambient environment of the mobile robot. The processing circuitry determines path indication information according to the sensing data. The path indication information includes a planned path of the mobile robot. The processing circuitry projects the path indication information on a target projection surface.
-
公开(公告)号：US11238605B2
公开(公告)日：2022-02-01
申请号：US16931155
申请日：2020-07-16
Inventor： Chao Chen
Abstract: Embodiments of this application disclose a method, a computer device, and a storage medium for simultaneous localization and mapping, applied to the field of information processing technologies. In the simultaneous localization and mapping method in the embodiments, edge information of a current frame image captured by an image capturing apparatus is obtained, and the edge information is divided into structural features of a plurality of first components, where each first component may correspond to a partial structure of one object in the current frame image, or correspond to at least one object. Then, a correspondence between first components in the current frame image and second components in a reference frame image may be obtained by matching the structural features of the first components with structural features of the second components in the reference frame image. Finally, the image capturing apparatus may be simultaneously localized according to the correspondence.
-
-
-
-
-
Search Results
6
records
(took 0.02 seconds)
