公开(公告)号：WO2023279936A1

公开(公告)日：2023-01-12

申请号：PCT/CN2022/099312

申请日：2022-06-17

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： YI, Zili ,  TANG, Qiang ,  RAMIYA SRINIVASAN, Vishnu Sanjay ,  XU, Zhan

IPC: G06T3/00

Abstract: Methods and systems for high-resolution image manipulation are disclosed. An original high-resolution image to be manipulated is obtained, as well as a driving signal indicating a manipulation result. The original high-resolution image is down-sampled to obtain a low-resolution image to be manipulated. Using a trained manipulation generator, a low-resolution manipulated image and a motion field are generated from the low-resolution image. The motion field represent pixel displacements of the low-resolution image to obtain the manipulation indicated by the driving signal. A high-frequency residual image is computed from the original high-resolution image. A high-frequency manipulated residual image is generated using the motion field. A high-resolution manipulated image is outputted by combining the high-frequency manipulated residual image and a low-frequency manipulated image generated from the low-resolution manipulated image by up-sampling.

公开(公告)号：WO2023083231A1

公开(公告)日：2023-05-19

申请号：PCT/CN2022/130995

申请日：2022-11-10

Applicant: HUAWEI TECHNOLOGIES CO., LTD.

Inventor： TANG, Qiang ,  RAMIYA SRINIVASAN, Vishnu Sanjay ,  CHEN, Shao Hua ,  XU, Zhan

IPC: G06N3/08 ,  G06T7/11

Abstract: This disclosure provides for methods and a system for multiple instance segmentation and tracking. According to an aspect a method is provided. The method includes sending an image (110) to a backbone network (102) and generating image feature outputs. The method further includes sending the image feature outputs to a spatial attention module (122) for generating a feature map associated with objects in the image (110). The method further includes sending the feature map to a category feature module (124) for generating an instance category output (126) indicating the objects. The method further includes sending the image feature outputs to a mask generating module for generating masks (146). The method further includes generating: the instance category output (126) via the category feature module (124), and the masks (146) via the mask generating module. In some embodiments, the method further includes generating re-identification embedding information (134) associated with the objects based on image feature outputs.