Table 3.
Summary of detection results of different models on KITTI object detection challenge.
| Methods | car (e) |
car (m) |
car (h) |
pede strian (e) |
pede strian (m) |
pede strian (h) |
cyclist (e) |
cyclist (m) |
cyclist (h) |
Number of parameters |
|---|---|---|---|---|---|---|---|---|---|---|
| RRC | — | 89.85 | — | — | 75.33 | — | — | 76.47 | — | 148M |
| MS-CNN | 90.03 | 89.02 | 76.11 | 83.92 | 73.70 | 68.31 | 84.06 | 75.46 | 66.07 | 551M |
| SqueezeDet | 90.2 | 84.7 | 73.9 | 77.1 | 68.3 | 65.8 | 82.9 | 75.4 | 72.1 | 2M |
| SqueezeDet+ | 90.4 | 87.1 | 78.9 | 81.4 | 71.3 | 68.5 | 87.6 | 80.3 | 78.1 | 6.7M |
| fire-FRD-CNN (ours) | 93.1 | 88.4 | 79.7 | 84.2 | 76.8 | 69.2 | 87.9 | 83.6 | 79.6 | 14M |