计算机视觉领域的Transformer系列算法  

Transformer Series Algorithms in the Field of Computer Vision

作　　者：张珈睿 吴超[1] ZHANG Jiarui;WU Chao(Tianjin University of Science and Technology,Tianjin 300457,China)

机构地区：[1]天津科技大学,天津300457

出　　处：《信息与电脑》2023年第2期101-103,共3页Information & Computer

摘　　要：目标检测是当前人工智能领域最火热的研究方向,而研究目标检测问题最重要的方法是深度学习。学者对特征提取、数据处理、网络结构及损失函数等方面进行了很多研究,提出了很多解决目标检测问题的算法,如Transformer系列算法。DETR (Detection Transformer)算法和可变形DETR算法通过集合预测损失实现了真正意义上的端到端的目标检测方法,其算法性能更是超越了传统算法。Target detection is the hottest research direction in the field of artificial intelligence,and the most important method to study target detection is deep learning.Scholar have studied feature extraction,data processing,network structure,loss function and other aspects,and put forward many algorithms for target detection problems,such as Transformer series algorithms.The DETR(Detection Transformer)algorithm and the deformable DETR algorithm realize the real end-to-end target detection method through the set prediction loss,and their algorithm performance is even better than the traditional algorithm.

关 键 词：计算机视觉 TRANSFORMER 算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]