Challenge and Dataset on Large-scale Human-centric Video Analysis in Complex Events (HiEve)

Short Name: RTU++
Complete name: EMNT with RTU++
Description:

The newest results of EMNT.

Runtime:

23

Submissions:

4

Benchmark
performance:

MOTA w_MOTA IDF1 MT ML FP FN ID_Sw ID_Sw_DT Frag MOTP
60.1363 55.2540 67.0105 32.0042 28.2267 1591 25586 598 85 693 80.7424

Detailed
performance:

Video ID MOTA w_MOTA IDF1 MT ML FP FN ID_Sw ID_Sw_DT Frag MOTP
20 89.3797 87.3666 92.2353 77.5510 6.1224 62 1191 14 6 26 83.4624
21 38.2851 34.2851 45.3280 13.3333 30.0000 49 546 24 1 16 72.7823
22 36.0874 31.1476 40.4340 18.4466 35.9223 225 5814 165 12 184 75.6843
23 91.1686 86.7013 87.0588 78.5714 14.2857 86 299 11 5 28 79.2829
24 20.4303 9.1039 29.1997 8.8146 55.9271 225 10306 156 38 166 76.6570
25 65.1186 59.8528 78.5887 36.7816 0.5747 528 1533 57 8 48 83.1411
26 62.4731 60.3231 56.0579 52.9412 29.4118 75 1300 21 2 28 80.2819
27 82.4247 76.9129 78.7380 67.7419 8.0645 145 703 45 7 39 81.8262
28 68.7722 62.4556 63.3490 30.0000 30.0000 86 691 14 4 36 76.3426
29 68.9763 67.1721 68.8584 56.0000 16.0000 30 641 23 1 41 80.3492
30 60.9178 60.9178 70.6394 28.8889 2.2222 6 750 19 0 17 82.1466
31 87.5681 87.5681 92.8419 88.8889 4.4444 34 237 3 0 5 81.6894
32 68.6011 67.8454 67.1310 40.0000 25.0000 40 1575 46 1 59 81.7154