Figure 1.
Network structure of YOLO v4 algorithm
TABLE I.
Environment configuration
| Hardware environment | processor Graphics card | Intel(R) XEON W-2133 Nvidia TITAN XP 12G |
|---|---|---|
| Software Environment | operating system | Ubuntu 16.04 |
| Deep learning framework | Tensorflow-gpu | |
| Programming language | Python | |
| translater | Pycharm2019.1 |
Figure 2.
Experimental framework
Figure 3.
Annotation of car instance segmentation
Figure 4.
The json file of the car label
Figure 5.
The txt file of the image tag
Figure 6.
Add Gaussian noise
Figure 7.
Median fuzzy processing
Figure 8.
Object number before and after data amplification
TABLE II.
Main network parameter values
| Parameter | Value | Parameter | Value |
|---|---|---|---|
| LEARN_RATE_INIT | 1e-4 | MOVING_AVE_DECAY | 0.9999 |
| LEARN_RATE_END | 1e-6 | STAGE_EPOCHS | 100 |
TABLE III.
Comparison of test results
| AP(%) | Car | Bus | Person | Motorbike | Bicycle | mAP(%) |
|---|---|---|---|---|---|---|
| yolo v4 | 0.98 | 0.93 | 0.92 | 0.81 | 0.51 | 82.95 |
| Improved yolo v4 | 0.99 | 0.93 | 0.92 | 0.81 | 0.58 | 84.98 |
Figure 9.
Learning rate change curve
Figure 10.
Experimental results based on Yolo v4
Figure 11.
Experimental results based on improved YOLO v4