计算智能和神经科学
研究文章

Multibranch对象检测方法对交通场景

表3

比较各种方法的平均精度(美联社)(%)。
模型 阈值
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65
RCNN [4] 37.32 38.55 40.29 42.78 43.46 44.12 46.29 47.04 51.12 53.47 57.78 59.04
快RCNN [9] 41.92 42.47 43.05 45.11 47.38 48.61 50.29 52.33 55.69 59.61 65.29 68.33
SSD (15] 47.74 49.07 52.14 54.97 57.61 60.14 62.24 64.39 65.57 68.14 71.24 74.39
面具RCNN [13] 51.45 53.23 55.01 56.26 58.38 59.21 62.88 64.49 67.13 70.21 74.44 77.27
SINet [2] 55.41 56.65 57.10 60.21 65.44 68.43 69.98 72.12 74.56 79.28 81.36 82.43
YOLOv3 [17] 54.94 55.07 57.34 59.30 62.35 64.82 66.32 68.73 69.18 72.82 75.32 79.73
MBNet网 58.25 59.31 61.45 62.67 64.94 67.19 70.01 72.33 75.12 77.19 80.01 83.68