Author Affiliations
School of Instrument Science and Engineering, Southeast University, Nanjing, Jiangsu 210096, Chinashow less
Fig. 1. Relative parameters of Darknet-53
Fig. 2. Distribution of traffic light label
Fig. 3. Result of clustering analysis
Fig. 4. Optimized YOLOv3 network structure
Fig. 5. Average loss function curve of optimized YOLOv3 algorithm
Fig. 6. Average IOU curve of optimized YOLOv3 algorithm
Fig. 7. Detection results of traffic lights by each network. (a)(c) Detection results of YOLOv3 network; (b)(d) detection results of optimized YOLOv3 network
Fig. 8. AP curve of YOLOv3 network
Fig. 9. AP curve of optimized YOLOv3 network
Fig. 10. Comparison of average precision under different scenes and object sizes
Fig. 11. Comparison of detection results. (a) Detection results of YOLOv3 algorithm; (b) detection results of optimized YOLOv3 algorithm
k=4 | k=5 | k=6 | k=7 | k=8 | k=9 |
---|
(5, 12) | (4, 12) | (4, 12) | (3, 12) | (4, 10) | (3, 11) | (5, 21) | (5, 20) | (6, 18) | (5, 18) | (5, 16) | (5, 13) | (8, 18) | (9, 27) | (6, 11) | (6, 10) | (7, 12) | (5, 21) | (13, 39) | (7, 14) | (7, 30) | (7, 17) | (7, 22) | (7, 12) | | (14, 47) | (12, 21) | (8, 33) | (8, 33) | (8, 18) | | | (14,47) | (12, 21) | (12, 46) | (8, 31) | | | | (15, 49) | (13, 22) | (12, 45) | | | | | (19, 56) | (14, 23) | | | | | | (19, 57) |
|
Table 1. Clustering results under different k values
Algorithm | Time /ms | Speed /(frame·s-1) | Precision /% | Recall /% | AP /% |
---|
YOLOv3 | 37 | 27 | 66.51 | 49.28 | 37.67 | Optimized YOLOv3 algorithm | 33 | 30 | 69.70 | 57.58 | 46.78 |
|
Table 2. Comparison of total object detection results
Network | Big object | Small object |
---|
P /% | R /% | AP /% | P /% | R /% | AP /% |
---|
YOLOv3 | 83.95 | 69.32 | 66.53 | 54.59 | 37.93 | 24.00 | Optimized YOLOv3 | 86.71 | 75.68 | 72.54 | 56.58 | 44.48 | 29.84 |
|
Table 3. Detection precision, recall, and average precision under different object sizes
Network | Night scene | Daytime scene | Rain scene | Non-rain scene |
---|
P/% | R/% | AP/% | P/% | R/% | AP/% | P/% | R/% | AP/% | P/% | R/% | AP/% |
---|
YOLOv3 | 65.36 | 49.98 | 37.67 | 70.45 | 44.04 | 36.15 | 70.28 | 51.00 | 41.18 | 63.71 | 47.29 | 34.77 | Optimized YOLOv3 | 68.32 | 57.24 | 44.58 | 72.06 | 59.11 | 49.96 | 73.79 | 58.86 | 49.89 | 70.01 | 57.10 | 45.15 |
|
Table 4. Detection precision, recall, and average precision under different scenes