Plant Image Recognition with Complex Background Based on Effective Region Screening

Xiaoyu Song; Liting Jin; Yang Zhao; Yue Sun; Tong Liu

doi:10.3788/LOP57.041016

Laser & Optoelectronics Progress, Volume. 57, Issue 4, 041016(2020)

Plant Image Recognition with Complex Background Based on Effective Region Screening

Xiaoyu Song^1、**, Liting Jin^1、*, Yang Zhao², Yue Sun¹, and Tong Liu¹

Author Affiliations

¹School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou, Gansu 730070, China

²Department of Information Engineering, Longqiao College of Lanzhou University of Finance and Economics, Lanzhou, Gansu 730101, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Structure of CNN

Download full size

Fig. 2. Principle diagram of Mask R-CNN effective region screening

Download full size

Fig. 3. Partial data display of open dataset. (a) Dataset partial pictures of Oxford 102 Flowers; (b) dataset partial pictures of Flavia

Download full size

Fig. 4. Number of different types of plants in GLT datasets

Download full size

Fig. 5. Flow chart of MRC-GoogleNet model

Download full size

Fig. 6. Plant image recognition model based on effective region screening

Download full size

Fig. 7. Accuracy versus Epoch. (a) AlexNet model; (b) GoogleNet model; (c) MRC-GoogleNet model

Download full size

Fig. 8. Loss value versus Epoch. (a) AlexNet model; (b) GoogleNet model; (c) MRC-GoogleNet model

Download full size

Table 1. Parameters of screening model CNN

View table

Table 1. Parameters of screening model CNN

Network layer	Input	Filter	Stride	Output
Input	32×32×1			32×32×1
Conv 1	32×32×1	5×5×16	1	28×28×6
Max Pool 1	28×28×6	2×2	2	14×14×6
Conv 2	14×14×6	5×5×16	1	10×10×16
Max Pool 2	10×10×16	2×2	2	5×5×16
FC1	5×5×16			120
FC2				84
FC3				3

Table 2. Training parameters of GoogleNet model
View table
Table 2. Training parameters of GoogleNet model
Parameter Value
Initial learning rate(α) 0.001
Rate of decline in learning rate(β) 0.96
Weight decay 0.0002
Momentum 0.9
Batch size 16
Number of iterations for learning rate reduction 20000

Table 3. Comparison results of model accuracy
View table
Table 3. Comparison results of model accuracy
Model Training time /s Accuracy /%
AlexNet 6482 77.85
GoogleNet 9371 84.32
MRC-GoogleNet 12986 95.21

Table 4. Recognition accuracy of different plant species

View table

Table 4. Recognition accuracy of different plant species

Plant species	AlexNet	GoogleNet	MRC-GoogleNet
Trifolium pratense	0.762	0.829	0.948
Dandelion	0.793	0.857	0.935
Hydrocleys nymphoides	0.837	0.891	0.986
Viola philippica	0.824	0.887	0.925
Datura stramonium	0.808	0.875	0.939
Portulaca oleracea	0.739	0.805	0.943
Nomocharis pardanthina	0.847	0.904	0.978
Clerodendrum thomsoniae	0.745	0.808	0.897
Uncarina grandidieri	0.773	0.859	0.954
Scabiosa comosa	0.759	0.819	0.899
Arctium lappa	0.749	0.814	0.958
Chelidonium majus	0.757	0.829	0.947
Trientalis europaea	0.795	0.863	0.894
Bellis perennis	0.801	0.869	0.976
Primula malacoides	0.742	0.799	0.928
Glechoma longituba	0.731	0.781	0.966
Tropaeolum majus	0.789	0.853	0.981
Plumeria rubra' Acutifolia'	0.791	0.856	0.967
Clerodendrum bungei	0.766	0.838	0.963
Talinum paniculatum	0.761	0.815	0.959

Tools

Get Citation

Copy Citation Text

Xiaoyu Song, Liting Jin, Yang Zhao, Yue Sun, Tong Liu. Plant Image Recognition with Complex Background Based on Effective Region Screening[J]. Laser & Optoelectronics Progress, 2020, 57(4): 041016

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites