Attention Mechanism in Computer Vision

Preface This paper systematically introduces the different categories of the Attention mechanism, and introduces the principles, advantages and disadvantages of each category.

Welcome to the public account CV technical guide , focusing on the technical summary of computer vision, the latest technology tracking, the interpretation of classic papers, and CV recruitment information.

Overview

The purpose of the Attention mechanism is to focus on useful information and reduce the proportion of unimportant information. Attention mechanisms can be divided into 6 categories, including 4 basic categories and 2 combined categories. The four basic categories are channel attention, spatial attention, temporal attention, and branch attention. The two combination categories are the combination of channel and space, and the combination of space and time.

Welcome to the public account CV technical guide , focusing on the technical summary of computer vision, the latest technology tracking, the interpretation of classic papers, and CV recruitment information.

CV Tech Guide creates a free knowledge planet . Follow the WeChat account of the official account to add editors and invites to join.

​​

The knowledge planet is under construction. Members who can write the following are welcome to contact me (WeChat: FewDesire, ID: As if there is light). Please note "research direction-location-ID" before adding

  1. A tutorial from TVM entry to practice

  2. TensorRT introductory to practice tutorial

  3. An introductory-to-practice tutorial for MNN

  4. Digital image processing and Opencv introductory to practice tutorial (must include digital image processing theory and opencv practice)

  5. OpenVINO entry-to-practice tutorial

  6. An introduction to libtorch tutorial

  7. Oneflow Getting Started to Practice Tutorial

  8. Detectron introductory to practice tutorial

  9. CUDA introductory to practice tutorial

  10. caffe source code reading

  11. pytorch source code reading

  12. Deep learning from entry to proficient (starting with convolutional neural networks)

  13. Interpretation of the latest summit. For example the recent CVPR2022 paper.

  14. Systematic review of various directions, development and evolution of main models, innovative ideas and advantages and disadvantages of each model, code analysis, etc. (Write it into multiple articles, it is not enough to write too simple, it is best to write an article with an important model)

  15. If you have something you want to write that is not mentioned above, you can contact me.

Guess you like

Origin juejin.im/post/7087429393718935589