摘要
Mask R-CNN作为一种优秀的实例分割算法,不仅能够对图像或视频中的每个感兴趣目标进行分类和定位,还能够对每个不同的目标进行分割。该算法在遥感图像智能解译,自动驾驶,智能医疗等计算机视觉领域具有极高的应用价值。因此如何对Mask R-CNN算法进行改进,提高实例分割的准确性,对计算机视觉领域的发展具有重要意义。文章对Mask R-CNN算法进行深入的分析和研究,提出了可用于Mask R-CNN算法的几个改进策略。通过实验验证,本文提出的改进策略对于提高Mask R-CNN算法的准确性具有一定的可行性。
As an excellent instance segmentation algorithm,Mask R-CNN can not only classification and location each target of interest in the image or video,but also segment each different target.The algorithm has extremely high application value in computer vision fields such as intelligent interpretation of remote sensing images,automatic driving,and intelligent medical treatment.Therefore,how to improve the Mask R-CNN algorithm to improve the accuracy of instance segmentation is of great significance to the development of computer vision.This paper conducts in-depth analysis and research on the Mask R-CNN algorithm,and proposes several improvement strategies that can be used for the Mask R-CNN algorithm.Through the experimental verification,the improved strategy proposed in this paper has certain feasibility for improving the accuracy of the Mask R-CNN algorithm.
作者
高爱
刘籽琳
GAO Ai;LIU Zi-lin(Institute Of Disaster Prevention,LangFang 065201,China)
出处
《电脑与信息技术》
2023年第1期47-49,共3页
Computer and Information Technology
基金
中央高校基本科研业务费大学生创新创业项目(项目编号:S202111775043)
中央高校基本科研业务费研究生创新项目(项目编号:ZY20220302)。
