Renjing Pei

Title: Renjing Pei and Vision-Transforming Technology.

Introduction

Renjing Pei is an inventor based in Shenzhen, China, and is associated with Huawei Technologies Co., Ltd. His patent work includes a computer-implemented approach to vision transforming that combines convolutional processing with multi-head self-attention.

Latest Patents

The provided information identifies one patent for Pei: “Computer-implemented methods, computing systems, and non-transitory machine-readable mediums for vision transforming.” The method divides image channels within image tiles into portions, processes one portion using depthwise convolution and another using multi-head self-attention, then combines the results. The combined representations are used to help identify objects in the image. Pei is credited with 1 patent in the supplied data.

Career Highlights

Pei’s listed invention addresses image understanding by bringing together two different processing techniques. Depthwise convolution processes local image features, while multi-head self-attention can capture relationships across image information. The method combines these outputs to support object identification.

Collaborations

The supplied information names Deepak Sridhar and Jizong Peng as Pei’s coworkers. It does not specify their roles in the patent or describe particular collaborative contributions.

Conclusion

Renjing Pei’s listed patent presents a hybrid approach to vision transforming that combines depthwise convolution and multi-head self-attention to support object identification.