Beijing Xiaomi Pinecone Electronics Co., Ltd.

Changxin Liu

Fei Wang

Pengfei Yu

Jiagao Hu

A video processing method includes: extracting a plurality of video frames from a source video with a first aspect ratio; clipping each of the plurality of video frames multiple times to generate a plurality of candidate clipping boxes with a second aspect ratio; respectively determining scores of the plurality of candidate clipping boxes of each of the plurality of video frames, the scores representing playing effects of the video frames in response to that each of the candidate clipping boxes is taken as a clipping range; selecting a clipping box of each of the video frames from the plurality of candidate clipping boxes based on the scores; and clipping each of the video frames by using the selected clipping box, and generating a target video with the second aspect ratio.

Video processing method, video processing device, and storage medium

The present disclosure relates to an image processing method that can include that an image to be processed is acquired, at least one category is selected via category information of the image to be processed, and a thermodynamic diagram for each category in the at least one category can be determined via the category information. The method can further include that a first positioning frame set corresponding to a target object in the image to be processed is respectively determined for the thermodynamic diagram for each category in the thermodynamic diagram for the at least one category, and a second positioning frame set of the image to be processed is determined according to an unsupervised target detection algorithm. Additionally, the method can include that a target positioning frame set in the image to be processed is determined according to the first positioning frame set and the second positioning frame set.

Growing community of inventors

Beijing, China

Changxin Liu