Antenna-PCI matching on Real-world 4G/5G dataset Overall
87.91Top-1 AccuracyVideo Swin (TER Framework)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Video Swin (TER Framework)Video Encoder=Video Swin, Training Paradigm=Pretrain + SFT2026.01 | 87.91 | 91.94 | |
| ViViT (TER Framework)Video Encoder=ViViT, Training Paradigm=Pretrain + SFT2026.01 | 79.87 | 88.24 | |
| TimeSformer (TER Framework)Video Encoder=TimeSformer, Training Paradigm=Pretrain + SFT2026.01 | 68.11 | 80.51 | |
| Video Swin (TER Framework)Video Encoder=Video Swin, Training Paradigm=End2End2026.01 | 62.12 | 89.93 | |
| ViViT (TER Framework)Video Encoder=ViViT, Training Paradigm=End2End2026.01 | 56.39 | 85.91 | |
| TimeSformer (TER Framework)Video Encoder=TimeSformer, Training Paradigm=End2End2026.01 | 30.36 | 64.29 |