| 
                Challenge and Dataset on Large-scale Human-centric Video Analysis in Complex Events (HiEve) | 
RetinaNet (Baseline)
none
1
Benchmark
					performance:
| MR | AP | JI | 
|---|---|---|
| 0.7041 | 0.6082 | 0.4898 | 
Detailed
performance:
| Video ID | MR | AP | JI | 
|---|---|---|---|
| 20 | 0.5268 | 0.9039 | 0.7480 | 
| 21 | 0.9174 | 0.4564 | 0.3459 | 
| 22 | 0.9556 | 0.4694 | 0.3746 | 
| 23 | 0.2148 | 0.9427 | 0.8086 | 
| 24 | 0.9557 | 0.2268 | 0.1946 | 
| 25 | 0.9976 | 0.0163 | 0.0390 | 
| 26 | 0.6562 | 0.5783 | 0.4585 | 
| 27 | 0.3958 | 0.8940 | 0.7625 | 
| 28 | 0.7371 | 0.6441 | 0.4715 | 
| 29 | 0.7621 | 0.6238 | 0.4463 | 
| 30 | 0.6222 | 0.7661 | 0.5843 | 
| 31 | 0.6141 | 0.8518 | 0.7138 | 
| 32 | 0.7984 | 0.5336 | 0.4197 |