Shadow Detection Method for CRC-RetinaNet Photovoltaic Panel Based on Multiscale Fusion

Jun Wu; Penghui Fan; Manli Wang

doi:10.3788/LOP202259.1615009

Laser & Optoelectronics Progress, Volume. 59, Issue 16, 1615009(2022)

Shadow Detection Method for CRC-RetinaNet Photovoltaic Panel Based on Multiscale Fusion

Jun Wu, Penghui Fan, and Manli Wang^*

Author Affiliations

School of Physics & Electronic Information Engineering, Henan Polytechnic University, Jiaozuo 454003, Henan , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(10)

Fig. 1. Structure of CRC-RetinaNet. (a) CSP backbone; (b) recursive-FPN; (c) class+box subnet

Download full size

Fig. 2. Schematic of CSP DenseNet

Download full size

Fig. 3. Cluster results of photovoltaic panel shadow dataset

Download full size

Fig. 4. Training loss decline curve

Download full size

Fig. 5. P-R curves of CRC-RetinaNet algorithm. (a) Photovoltaic panel; (b) photovoltaic panel shielding

Download full size

Fig. 6. Detection results。(a) (b) (c) (d) (e) Dense and overlapping targets; (d) (e) small targets

Download full size

Table 1. Confusion matrix of classification results
View table
Table 1. Confusion matrix of classification results
Ground truth Detection result
Positive Negative
Positive TP FN
Negative FP TN

Table 2. Dataset distribution
View table
Table 2. Dataset distribution
Class Number of training set and verification set Number of test set
PVP 46203 6017
PVP_shielding 46220 5049

Table 3. Performance comparison of different algorithms

View table

Table 3. Performance comparison of different algorithms

Algorithms	AP /%		mAP /%	FPS	Model size /MB
Algorithms	PVP	PVP_shielding	mAP /%	FPS	Model size /MB
M2Det	96.89	90.21	93.55	24.8	226.66
EfficientDet	86.29	39.09	62.69	30.7	16.16
SSD	55.07	30.55	42.81	41.4	100.40
SSD-anchor	63.23	52.15	57.69	42.3	100.40
Faster R-CNN	40.66	20.44	30.55	16.0	301.05
Faster R-CNN-anchor	97.68	94.42	96.05	18.5	301.05
YOLOv3	95.81	98.33	97.07	34.9	246.30
YOLOv4	96.12	98.09	97.10	30.2	266.30
RetinaNet	96.92	93.52	95.22	15.6	139.30
CRC-RetinaNet	99.33	99.15	99.24	32.4	66.26

Table 4. Comparison of detection accuracy of different models

View table

Table 4. Comparison of detection accuracy of different models

Group	CSP	RFPN	Mish +Leaky ReLU	CIoU loss	AP /%		mAP /%	FPS	Model size /MB
Group	CSP	RFPN	Mish +Leaky ReLU	CIoU loss	PVP	PVP_shielding	mAP /%	FPS	Model size /MB
1	×	×	×	×	96.92	93.52	95.22	15.6	139.30
2	√	×	×	×	98.29	97.12	97.71	33.5	45.98
3	√	√	×	×	98.92	98.17	98.55	32.1	66.26
4	√	√	√	×	99.06	98.88	98.97	31.6	66.26
5	√	√	√	√	99.33	99.15	99.24	32.4	66.26

Tools

Get Citation

Copy Citation Text

Jun Wu, Penghui Fan, Manli Wang. Shadow Detection Method for CRC-RetinaNet Photovoltaic Panel Based on Multiscale Fusion[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1615009

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites