notesum.ai

Published at November 6

UnityGraph: Unified Learning of Spatio-temporal features for Multi-person Motion Prediction

cs.CV

cs.AI

Released Date: November 6, 2024

Authors: Kehua Qu, Rui Ding, Jin Tang

Arxiv: http://arxiv.org/abs/2411.04151v1

Refer to caption

Methods

CMU-Mocap

(3 persons)

MuPoTS-3D

(2

\sim

3 persons)

Mix1

(9

\sim

15 persons)

Mix2

(11 persons)

1sec

2sec

3sec

1sec

2sec

3sec

1sec

2sec

3sec

1sec

2sec

3sec

LTD [66]

13.7

21.9

32.6

11.9

18.1

23.4

21.0

31.9

41.5

17.2

25.8

34.5

HRI [35]

14.9

26.0

30.7

9.4

16.8

22.9

18.0

31.4

42.1

16.0

27.1

36.7

MRT [50]

9.6

15.7

21.8

8.9

15.9

22.2

17.3

29.9

39.7

12.9

20.9

28.2

TCD* [10]

10.2

16.1

19.5

9.0

15.8

21.7

16.8

28.7

37.7

12.1

19.2

26.7

TBIFormer* [11]

8.0

13.4

19.0

8.7

15.1

20.9

14.0

24.4

31.2

12.3

18.2

26.1

JRT [52]

8.3

13.9

18.5

8.9

15.5

21.3

-

-

-

-

-

-

Ours

7.8

13.0

17.3

8.4

14.7

20.1

13.6

23.6

29.4

11.0

16.2

23.8