计算机科学 ›› 2023, Vol. 50 ›› Issue (6A): 220400023-5.doi: 10.11896/jsjkx.220400023
孙开伟, 王支浩, 刘虎, 冉雪
SUN Kaiwei, WANG Zhihao, LIU Hu, RAN Xue
摘要: 随着人工智能的发展,深度学习在计算机视觉研究中引起了广泛关注,在单目标跟踪领域开始对基于深度学习的单目标跟踪算法加以研究。深度学习算法的算法复杂度相对较高,将目标分类和目标状态估计完整的分割出来,有利于对每一个任务的深层探讨。但现阶段的单目标跟踪算法不能很好地应对复杂的跟踪环境,模型遇到复杂跟踪环境时,经常会跟踪到背景的某一块区域或者跟踪到周围的相似目标。为了解决以上问题,文中提出了一种方法,在目标分类和目标状态估计任务中分别加入了不同的注意力机制,使得模型能够更好地处理背景混乱和相似目标遮挡的情况。为了验证上述方法的有效性,文中在多个数据集上做了大量的对比实验,并且和之前的基于深度学习的单目标跟踪算法进行比较,所提算法在EAO指标上有了3.1%的提升,在Robustness指标上有了2.3%的提升,表明了其有效性和先进性。
中图分类号:
[1]ZHANG W C,SUN C M.Corner Detection Using Multi-directional Structure Tensor with Multiple Scales[J].International Journal of Computer Vision,2020,128(2):438-459. [2]ZHAO F,WANG J Q,WU Y,et al.Adversarial Deep Tracking[J].IEEE Transactions on Circuits & Systems for Video Technology,2018,29(7):1998-2011. [3]YU L Y,FAN C X,MING Y.Improved Target Tracking Algorithm based on Kernelized Correlation Filter[J].Journal of Computer Applications,2015,35(12):3550-3554. [4]WANG N,SONG Y B.Unsupervised Deep Tracking[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).Long Beach:IEEE,2020:1308-1317. [5]WU Y,LIM J.Object Tracking Benchmark[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1834-1848. [6]WU Y,LIM J.A Numerical Algorithm for the Coupled PDEs Control Problem[J].Computation Aleconomics,2019,53(2):697-707. [7]DANELLJAN M,BHAT G,KHAN F S,et al.ATOM:Accurate tracking by overlap maximization[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR).Long Beach:IEEE,2019:4655-4664. [8]KUMAR D,MONDAL S,KARUPPUSWAMI S,et al.Har-monic RFID Communication Using Conventional UHF System[J].IEEE Journal of Radio Frequency Identification,2019,3(4):227-235. [9]ZHANG X M,ZHANG X H,DU X D,et al.Learning multi-domain convolutional network for rgb-t visual tracking[C]//2018 11th International Congress on Image and Signal Processing,BioMedical Engineering and Informatics(CISP-BMEI).Beijing:IEEE,2018:1-6. [10]JUNG I,SON J,BAEK M,et al.Real-time mdnet[C]//Procee-dings of the European Conference on Computer Vision(ECCV).Cham:Springer,2018:83-98. [11]FERRAZ P A P,DE OLIVEIRA B A G,FERREIRA F M F,et al.Three-stage RGBD architecture for vehicle and pedestrian detection using convolutional neural networks and stereo vision[J].IET Intelligent Transport Systems,2020,14(10):1319-1327. |
|