Table 1 Comparison with other multispectral detectors on the KAIST dataset.
From: Research on pedestrian detection method based on multispectral intermediate fusion using YOLOv7
Method | Runtime (s) | Precision (%) | Recall (%) | F1-score (%) | mAP (%) | MR (%) |
|---|---|---|---|---|---|---|
ACF | 2.73 | 65.0 | 62.8 | 63.9 | 70.0 | 27.5 |
Halfway fusion | 0.43 | 70.5 | 68.9 | 69.7 | 74.0 | 23.0 |
Fusion RPN + BF | 0.8 | 74.8 | 72.5 | 73.6 | 78.5 | 18.0 |
IAF R-CNN | 0.21 | 78.2 | 76.9 | 77.6 | 82.0 | 16.0 |
IATDNN + IASS | 0.25 | 79.5 | 78.0 | 78.7 | 83.3 | 14.5 |
CIAN | 0.07 | 81.0 | 80.0 | 80.5 | 84.5 | 12.0 |
MSDS-RCNN | 0.22 | 81.8 | 81.0 | 81.4 | 85.0 | 10.5 |
ARCNN | 0.12 | 80.8 | 79.5 | 80.1 | 84.0 | 13.5 |
MLPD | 0.12 | – | – | – | – | 7.58 |
GAFF | 0.09 | – | – | – | 72.9 | 10.62 |
ICA-fusion | – | – | – | – | 79.2 | 7.17 |
OURS | 0.08 | 83.5 | 82.3 | 82.9 | 85.7 | 7.5 |
