International Business Machines Corporation

Intel Corporation

Beijing Baidu Netcom Science and Technology Co., Ltd.

Jia Wei

Yu Wang

Anbang Yao

Yurong Chen

Ping Li

Ping Hu

Dongqi Cai

Yang Zhang

Wei Peng

Du Chen

Yuzhen Chen

Jun J Zhang

Jing G Zhang

Zhenxing Cao

Heng Yao

Techniques are provided for neural network based, human attribute recognition, guided by anatomical key-points and statistic correlation models. Attributes include characteristics that can be visibly identified or inferred from an image, such as gender, hairstyle, clothing style, etc. A methodology implementing the techniques according to an embodiment includes applying an attribute feature extraction (AFE) convolutional neural network (CNN) to an image of a human to generate attribute feature maps based on the image. The method further includes applying a key-point guided proposal (KPG) CNN to the image of the human to generate proposed hierarchical regions of the image based on associated anatomical key-points. The method further includes generating recognition probabilities for the human attributes using a CNN combination layer that incorporates the attribute feature maps, the proposed hierarchical regions, and statistical correlation models (SCMs) which provide correlations between the features of the attribute feature maps and the proposed hierarchical regions.

Key-point guided human attribute recognition using statistic correlation models

A request for sharing information of a first user to a second user is received. Responsive to the request, information of the first user is obtained. The information is encrypted using a first server key to generate first encrypted information, the first server key being unavailable to the second user. The first encrypted information is encrypted using a second user key specific to the second user to generate second encrypted information for sharing with the second user.

Protected information sharing

A method retrieves in order a plurality of specific topic-based messages in a distributed publish/subscribe system, the distributed publish/subscribe system comprising at least one message transit server, the method comprising: determining an order of a plurality of to-be retrieved messages; recording, in message transit server sequence information, message transit servers where the plurality of messages are located, respectively, an order of the message transit servers in the message transit server sequence information corresponding to the determined order of the plurality of messages; and retrieving in order the plurality of messages from corresponding message transit servers using the message transit server sequence information.

Retrieving messages in order in a distributed publish/subscribe system

Growing community of inventors

Beijing, China

Jia Wei