Research Article
A Multibranch Object Detection Method for Traffic Scenes
| Submeter 2 | Model | Average | Crowded (daytime) | C | CP | P | B | BI | M | T |
| RCNN [4] | 43.94 | 42.12 | 38.28 | 40.32 | 55.17 | 41.31 | 44.14 | 46.25 | Faster RCNN [9] | 47.73 | 50.25 | 43.47 | 51.06 | 57.28 | 45.27 | 44.04 | 42.74 | SSD [15] | 58.00 | 59.27 | 51.32 | 62.21 | 66.57 | 57.22 | 58.18 | 51.21 | Mask RCNN [13] | 57.78 | 55.25 | 58.67 | 60.20 | 70.23 | 56.56 | 52.33 | 51.19 | SINet [2] | 64.95 | 66.58 | 57.57 | 65.35 | 73.33 | 61.01 | 65.59 | 65.20 | YOLOv3 [17] | 60.03 | 63.63 | 55.72 | 58.34 | 68.63 | 58.01 | 56.70 | 59.21 | MBNet | 66.45 | 65.78 | 61.06 | 63.74 | 76.59 | 65.70 | 66.81 | 65.46 |
|
|