Fig. 1. Framework of Siamese network with channel spatial coordination attention module

Fig. 2. Convolutional blocks of MobileNetV2

Fig. 3. Convolution process abstract graph of MobileNetV2

Fig. 4. Grad-CAM network visualization results. (a) No attention module; (b) with attention module

Fig. 5. Channel spatial coordination attention module

Fig. 6. Success rate comparison of different attention module combinations on OTB-2015

Fig. 7. Precision comparison of different attention module combinations on OTB-2015

Fig. 8. Qualitative results of the nine tracking algorithms on different video sequences. (a) car6_5; (b) car17; (c) person9; (d) person1_s; (e) uav4; (f) wakeboard6

Fig. 9. Results of problem sequence on uav1_1

Fig. 10. Results of the tracking algorithms on OTB-2015. (a) Success plot; (b) precision plot

Fig. 11. Tracking success plots of different attributes videos. (a) Scale variation; (b) aspect ratio change; (c) low resolution; (d) fast motion; (e) full occlusion; (f) partial occlusion; (g) out-of-view; (h) background clutter; (i) illumination variation; (j) viewpoint change; (k) camera motion; (l) similar object

Fig. 12. Tracking precision plots of different attributes videos. (a) Scale variation; (b) aspect ratio change; (c) low resolution; (d) fast motion; (e) full occlusion; (f) partial occlusion; (g) out-of-view; (h) background clutter; (i) illumination variation; (j) viewpoint change; (k) camera motion; (l) similar object

Fig. 13. Quantitative analysis of some video sequences. (a) Scale variation and aspect ratio change; (b) CLE

Table 1. Architecture of Siamese network based on MobieleNetV2