Object Detection Model Based on Multi-Scale Feature Integration

Wanjun Liu; Feng Wang; Haicheng Qu

doi:10.3788/LOP56.231007

Journals >Laser & Optoelectronics Progress >Volume 56 >Issue 23 >Page 231007 > Article

Laser & Optoelectronics Progress
Vol. 56, Issue 23, 231007 (2019)

Object Detection Model Based on Multi-Scale Feature Integration

Wanjun Liu, Feng Wang^*, and Haicheng Qu

Author Affiliations

College of Software, Liaoning Technical University, Huludao, Liaoning 125105, China

show less

DOI: 10.3788/LOP56.231007 Cite this Article Set citation alerts

Wanjun Liu, Feng Wang, Haicheng Qu. Object Detection Model Based on Multi-Scale Feature Integration[J]. Laser & Optoelectronics Progress, 2019, 56(23): 231007 Copy Citation Text

show less

Fig. 1. Flowchart of RF-YOLOv2 detection

Download full size

Fig. 2. Object function change curve

Download full size

Fig. 3. Residual block structure

Download full size

Fig. 4. Feature pyramid network

Download full size

Fig. 5. Flowchart of RF-YOLOv2

Download full size

Fig. 6. Number of categories appearing on KITTI data set

Download full size

Fig. 7. Loss graph for two models

Download full size

Fig. 8. Precision-Recall curves of two models. (a)(c)(e) YOLOv2 model;(b)(d)(f) RF-YOLOv2 model

Download full size

Fig. 9. Detection results. (a)(c)(e)(g)(i) Detection results of YOLOv2 model; (b)(d)(f)(h)(j) detection results of RF-YOLOv2 model

Download full size

Layerblock	Type	Numberof filters	Size /stride	Output
	Convolutional	32	3×3	416×416
	Maxpool		2×2/2	208×208
	Convolutional	64	3×3	208×208
1×	Convolutional	32	1×1
	Convolutional	64	3×3
	Residual			208×208
	Maxpool		2×2/2	104×104
	Convolutional	128	3×3	104×104
2×	Convolutional	64	1×1
	Convolutional	128	3×3
	Residual			104×104
	Maxpool		2×2/2	52×52
	Convolutional	256	3×3	52×52
4×	Convolutional	128	1×1
	Convolutional	256	3×3
	Residual			52×52
	Maxpool		2×2/2	26×26
	Convolutional	512	3×3	26×26
4×	Convolutional	256	1×1
	Convolutional	512	3×3
	Residual			26×26
	Maxpool		2×2/2	13×13
	Convolutional	1024	3×3	13×13
4×	Convolutional	512	1×1
	Convolutional	1024	3×3
	Residual			13×13
	Avgpool		Global	3
	Softmax

Table 1. RF-YOLOv2 network structure

Model	Accuracyofcar /%	Accuracy ofpedestrian /%	Accuracy ofcyclist /%	Detectionspeed /(frame·s^-1)
YOLOv2	68.56	44.26	55.95	46.4
RF-YOLOv2	87.88	52.91	74.05	30.3
YOLOv3	89.34	60.93	83.94	23.1

Table 2. Comparison of accuracy and detection speed

Number oftraining	RF-YOLOv2 model		YOLOv2 model
Number oftraining	Recallrate /%	I_OU /%	Recallrate /%	I_OU /%
10000	50.36	43.29	48.18	43.42
20000	55.45	46.34	53.11	45.98
30000	61.47	50.65	55.83	47.79
40000	64.92	52.56	54.13	46.72
50000	65.87	53.63	57.98	49.04

Table 3. Change process of recall rate and I_OU

Model	Accuracy of easy sample /%	Accuracy of moderate sample /%	Accuracy of hard sample /%
YOLOv2	70.56	57.32	50.44
Faster-rcnn	87.90	79.11	70.19
RF-YOLOv2	91.01	81.26	72.41

Table 4. Three sample detection results of car category

Model	Accuracy of easy sample /%	Accuracy of moderate sample /%	Accuracy of hard sample /%
YOLOv2	59.97	49.05	44.91
Faster-rcnn	78.35	65.91	61.19
RF-YOLOv2	64.35	57.02	53.94

Table 5. Three sample detection results of pedestrian category

Model	Accuracy of easy sample /%	Accuracy of moderate sample /%	Accuracy of hard sample /%
YOLOv2	56.47	56.68	53.02
Faster-rcnn	71.41	62.81	55.44
RF-YOLOv2	79.76	74.68	72.41

Table 6. Three sample detection results of cyclist category

Wanjun Liu, Feng Wang, Haicheng Qu. Object Detection Model Based on Multi-Scale Feature Integration[J]. Laser & Optoelectronics Progress, 2019, 56(23): 231007

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information