Table 8.
Detection performance of Mixed YOLOv3-LITE (832 × 832) for each category using VisDrone2018-Det validation dataset.
| Class | Images | Instances | Precision (%) | Recall (%) | F1 (%) | mAP (%) |
|---|---|---|---|---|---|---|
| awning-tricycle | 548 | 532 | 26.12 | 13.16 | 17.50 | 6.24 |
| bicycle | 548 | 1287 | 20.08 | 19.35 | 19.71 | 7.92 |
| bus | 548 | 251 | 47.60 | 47.41 | 47.50 | 40.87 |
| car | 548 | 14,064 | 61.36 | 76.54 | 68.11 | 70.79 |
| motor | 548 | 5125 | 44.08 | 43.61 | 43.85 | 32.74 |
| pedestrian | 548 | 8844 | 34.57 | 45.66 | 39.35 | 34.50 |
| people | 548 | 4886 | 42.32 | 35.45 | 38.58 | 23.39 |
| tricycle | 548 | 1045 | 33.38 | 24.69 | 28.38 | 15.25 |
| truck | 548 | 750 | 33.81 | 31.47 | 32.60 | 21.94 |
| van | 548 | 1975 | 48.55 | 40.71 | 44.29 | 31.31 |
| overall | 548 | 38,759 | 39.19 | 37.80 | 37.99 | 28.50 |