Table 3 Comparison of efficiency with state-of-the-art methods on the datasets of MOT2015 and MOT2016.

From: Multi-target detection and tracking based on CRF network and spatio-temporal attention for sports videos

Model

Datasets

MOT2015

MOT2016

Parameters(M)

Flops(G)

Inference Time(ms)

Trainning Time(s)

Parameters(M)

Flops(G)

Inference Time(ms)

Trainning Time(s)

KFC29

534.94

5.12

8.92

580.34

497.03

5.88

9.33

552.09

Shay et al.30

682.34

8.68

11.97

741.60

687.97

7.74

10.69

659.83

Yong et al.31

786.95

5.87

7.97

735.75

753.00

7.51

7.96

409.45

Fang et al.32

612.04

8.15

10.38

718.84

703.20

8.53

13.04

666.87

HDT network33

427.95

5.16

7.17

433.39

411.39

5.09

6.75

489.33

L-YOLOv434

339.64

3.55

5.35

325.12

317.18

3.66

5.60

335.62

Ours

336.61

3.53

5.34

321.37

310.79

3.36

5.12

329.26

Back to article page