Micro-gesture recognition on iMiGUE MiGA Track-1 (test)
74.66AccuracyXInsight Lab (PoseConv3D+VideoSwinT+SMILE)
Evaluation results
| Method | Links | |
|---|---|---|
| XInsight Lab (PoseConv3D+VideoSwinT+SMILE)Rank=MiGA’26 1st, Modality=J+L+R+D2026.06 | 74.66 | |
| Team YUVRank=MiGA’26 2nd2026.06 | 74.24 | |
| HFUT-VUT (PoseConv3D+VideoSwinT)Rank=MiGA’25 1st, Modality=J+L+R+T+F+D2026.06 | 73.21 | |
| Team xdRank=MiGA’26 3rd2026.06 | 71.37 | |
| HFUT-VUT (PoseConv3D)Rank=MiGA’24 1st, Modality=J+L+R2026.06 | 70.25 | |
| NPU-MUCIS (Res2Net3D+GCN)Rank=MiGA’24 2nd, Modality=J+R2026.06 | 70.19 | |
| awuniverse (ViT-Base+GAIE)Rank=MiGA’25 2nd, Modality=R2026.06 | 68.7 | |
| AIMRank=MiGA’26 4th, Core Methodology=Decoupled ST-CNN + PoseC3D + Swin3D & R(2+1)D, Modality=J+H+R2026.06 | 68.13 | |
| Lonelysheep (PoseConv3D)Rank=MiGA’25 3rd, Modality=J+L2026.06 | 67.01 | |
| HFUT-VUT (PoseConv3D)Rank=MiGA’23 1st, Modality=J+L2026.06 | 64.12 | |
| NPU-Stanford (Hyperformer)Rank=MiGA’23 2nd, Modality=J2026.06 | 63.02 |