面向无人驾驶的零样本记忆感知选择视觉跟踪模型

李杰,汪诗敏,王长城,崔亚峰,汪俊杰,周惟嘉,胡铮,兰海,杜玲,高猛

Zero-shot memory-aware selection visual tracking model for unmanned driving

Jie LI,Shimin WANG,Changcheng WANG,Yafeng CUI,Junjie WANG,Weijia ZHOU,Zheng HU,Hai LAN,Ling DU,Meng GAO

表 1 不同数据集上各算法的视觉目标跟踪结果对比

Tab.1 Comparison of visual object tracking results of various algorithms on different datasets

算法 LaSOT GOT-10k OTB100 $ {P}_{\mathrm{n}\mathrm{o}\mathrm{r}\mathrm{m}} $/% AUC/% P/% AO/% $ {\mathrm{O}\mathrm{P}}_{0.50} $/% $ {\mathrm{O}\mathrm{P}}_{0.75} $/% $ {S}_{\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{e}} $/% AUC/% P/% HIPTrack-B384[28] 82.9 72.7 79.5 77.4 88.0 74.5 79.2 71.0 80.2 AQATrack-B256[29] 81.9 71.4 78.6 73.8 83.2 72.1 76.4 72.8 83.1 ODTrack-B384[30] 83.2 73.2 80.6 77.0 87.9 75.1 75.6 73.0 81.8 LoRAT-B224[31] 80.9 71.7 77.3 72.1 84.9 75.0 80.4 72.3 82.5 OSTrack384[15] 81.1 71.1 77.6 73.7 83.2 70.8 77.6 55.9 75.8 SiamPRN++[32] 56.9 49.6 49.1 51.7 61.6 32.5 72.9 69.2 77.6 DiMP288[33] 64.1 56.3 56.0 61.1 71.7 49.2 66.4 74.3 78.4 Zero-shot 82.7 75.3 82.8 79.6 90.8 79.2 81.7 74.8 84.9