Alibaba Group Holding Limited

Zhejiang University

Fan Wu

Yang Li

Zhijun Yuan

Dong Liu

Yongying Yang

Huiting Chai

Pin Cao

A type identification is firstly performed on a to-be-processed access request, and when the to-be-processed access request is identified as a first-type access request, anomaly identification is then performed on the to-be-processed access request by using a machine learning model. The techniques of the present disclosure not only accurately identify an abnormal access request, but also effectively reduce the number of access requests that need to be identified by the machine learning model, thus saving computing resources of the device and improving the operating performance of the device.

Data processing method, device and storage medium

Upon receipt of a message including text content, matching is performed using a preset activity matching template. If the matching is successful, an activity card of a corresponding activity type is generated, and a related service application is displayed when the activity card is displayed. The present disclosure enables a user to see various related activity cards only by entering a list of activities without manually searching for, from a great number of message including text contents, a message including text content relevant to an activity. Moreover, the present disclosure enables the user to see the most related service application only by displaying an activity card without searching for and downloading an associated application program and without manually searching for an application relevant to an activity from hundreds of applications at a mobile terminal.

Text message processing

A defects evaluation system and method are provided in the present invention. Based on the principle of the microscopic scattering dark-field imaging, the present invention implements a sub-aperture scanning for the surface of spherical optical components and then obtains surface defects information with image processing. Firstly, the present invention takes full advantage of the characteristic that the surface defects of spherical optical components can generate scattering light when an annular illumination beam irradiates on the surface, to implement the sub-aperture scanning and imaging that covers the entire spherical surface. Then, a series of procedures such as the global correction of sub-apertures, the 3D stitching, the 2D projection and the digital feature extraction are taken to inspect spherical surface defects. Finally, actual size and position information of defects are evaluated quantitatively with the defects calibration data. The present invention achieves the automatic quantitative evaluation for surface defects of spherical optical components, which considerably enhance the efficiency and precision of the inspection, avoiding the influence of subjectivity on the results. Eventually, reliable numerical basis for the use and process of spherical optical components is provided.

Surface defects evaluation system and method for spherical optical components

Growing community of inventors

Zhejiang, China

Fan Wu