Combined Auxiliary Networks and Bird’s Eye View Method for Real-Time Multicategory Object Recognition

Table 2

Results of comparison trained/tested by the KITTI dataset.

Category	AP (regression); IoU > 0.5	AP (regression); IoU > 0.7	AP (category)	Lpoint	Lbox	sin θ/cos θ
Vehicle	90.0	77.4	97.7
Pedestrian	66.5	42.2	88.6
Cyclist	68.2	32.5	73.2
Vehicle	95.6	93.0	99.2			✓
Pedestrian	77.6	61.8	96.0			✓
Cyclist	85.1	68.3	83.1			✓
Vehicle	94.2	91.2	99.3	✓		✓
Pedestrian	81.4	63.2	95.3	✓		✓
Cyclist	86.7	79.2	87.8	✓		✓
Vehicle	95.7	92.2	99.1	✓	✓	✓
Pedestrian	83.3	64.3	95.9	✓	✓	✓
Cyclist	88.3	78.3	89.1	✓	✓	✓