Table 6.
The results of the proposed model on the TuSimple dataset when attention gate is embedded into different stages in the encoding network.
| Layer | ACC (%) | PRE | REC | F1-Measure |
|---|---|---|---|---|
| Up1 | 98.15 | 0.8691 | 0.9462 | 0.9060 |
| Up2 | 98.10 | 0.8683 | 0.9494 | 0.9072 |
| Up3 | 97.98 | 0.8686 | 0.9507 | 0.9088 |