J Arid Land (2022) 14(12): 1440–1455

https://doi.org/10.1007/s40333-022-0086-9

Science Press Springer-Verlag

Image recognition and empirical application of desert

plant species based on convolutional neural network
LI Jicai1, SUN Shiding2, JIANG Haoran2, TIAN Yingjie1,3,4*, XU Xiaoliang5
1
School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190, China;
2
School of Mathematical Sciences, University of Chinese Academy of Sciences, Beijing 100049, China;
3
Key Laboratory of Big Data Mining and Knowledge Management, Chinese Academy of Sciences, Beijing 100190, China;
4
Research Center on Fictitious Economy and Data Science, Chinese Academy of Sciences, Beijing 100190, China;
5
Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China

Abstract: In recent years, deep convolution neural network has exhibited excellent performance in
computer vision and has a far-reaching impact. Traditional plant taxonomic identification requires high
expertise, which is time-consuming. Most nature reserves have problems such as incomplete species
surveys, inaccurate taxonomic identification, and untimely updating of status data. Simple and accurate
recognition of plant images can be achieved by applying convolutional neural network technology to
explore the best network model. Taking 24 typical desert plant species that are widely distributed in the
nature reserves in Xinjiang Uygur Autonomous Region of China as the research objects, this study
established an image database and select the optimal network model for the image recognition of desert
plant species to provide decision support for fine management in the nature reserves in Xinjiang, such as
species investigation and monitoring, by using deep learning. Since desert plant species were not included in
the public dataset, the images used in this study were mainly obtained through field shooting and
downloaded from the Plant Photo Bank of China (PPBC). After the sorting process and statistical analysis,
a total of 2331 plant images were finally collected (2071 images from field collection and 260 images from
the PPBC), including 24 plant species belonging to 14 families and 22 genera. A large number of numerical
experiments were also carried out to compare a series of 37 convolutional neural network models with
good performance, from different perspectives, to find the optimal network model that is most suitable for
the image recognition of desert plant species in Xinjiang. The results revealed 24 models with a recognition
Accuracy, of greater than 70.000%. Among which, Residual Network X_8GF (RegNetX_8GF) performs the
best, with Accuracy, Precision, Recall, and F1 (which refers to the harmonic mean of the Precision and Recall
values) values of 78.33%, 77.65%, 69.55%, and 71.26%, respectively. Considering the demand factors of
hardware equipment and inference time, Mobile NetworkV2 achieves the best balance among the Accuracy,
the number of parameters and the number of floating-point operations. The number of parameters for
Mobile Network V2 (MobileNetV2) is 1/16 of RegNetX_8GF, and the number of floating-point
operations is 1/24. Our findings can facilitate efficient decision-making for the management of species
survey, cataloging, inspection, and monitoring in the nature reserves in Xinjiang, providing a scientific basis
for the protection and utilization of natural plant resources.

Keywords: desert plants; image recognition; deep learning; convolutional neural network; Residual Network X_8GF
(RegNetX_8GF); Mobile Network V2 (MobileNetV2); nature reserves

Citation: LI Jicai, SUN Shiding, JIANG Haoran, TIAN Yingjie, XU Xiaoliang. 2022. Image recognition and empirical
application of desert plant species based on convolutional neural network. Journal of Arid Land, 14(12): 1440–1455.
https://doi.org/10.1007/s40333-022-0086-9

∗
Corresponding author: TIAN Yingjie (E-mail: tyj@ucas.ac.cn)
Received 2022-09-01; revised 2022-10-14; accepted 2022-10-16
© Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Science Press and Springer-Verlag GmbH Germany,
part of Springer Nature 2022

http://jal.xjegi.com; www.springer.com/40333
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1441

1 Introduction
Wild plants constitute the main part of the ecosystem in the nature reserves. Thus, it is the
primary problem for managers to carry out species investigation and classification on the wild
plants (Li et al., 2020). Owing to the high professional knowledge demand, the traditional plant
classification and recognition is time-consuming and inefficient (Cao et al., 2018), the existing
species survey in most of nature reserves is not comprehensive, and the classification of plant
species is not accurate, leading to the problems such as status data update is not in time, and the
administration agencies of nature reserves are unable to make timely and effective protection and
management as well as countermeasures for ecological recovery (Liu et al., 2018; Wang et al.,
2019; Xiao, 2019). In June 2019, the General Office of the State Council of the People's Republic
of China issued a document to establish national park as the main body of the nature reserve
system, based on ecological environment regulation and big data platform, for use of information
technology means such as cloud computing and Internet of things for a comprehensive grasp of
nature reserve ecosystem composition, distribution, and dynamic change, thus providing a
scientific support for the management decisions of nature reserves (http://www.gov.cn/zhengce/
2019-06/26/content_5403497.htm). Therefore, how to use new information technology to obtain
plant-related data more efficiently and accurately has become an urgent problem to be solved by
researchers and managers.
With its powerful feature extraction, convolution neural network has significant advantages in
the recognition and analysis of high-dimensional data such as images, sounds, and texts, which
can reduce the damage of field specimens to fragile plant resources, decrease the difficulty of
identification and classification of similar plant species, and improve work efficiency (Mikolov et
al., 2011). In recent years, intelligent recognition of plant images has gradually become a research
hotspot (Liu, 2020). In the previous literature, convolutional neural network was used to
recognize the images of leaves, fruits, flowers, and other organs of plants under a simple
background (Hall et al., 2015; Abdullahi et al., 2017; Bargoti and Underwood, 2017; Coulibaly et
al., 2019; Cao et al., 2020). Researchers have classified and recognized the images of five crops
and 100 different ornamental plant species in different nature scenes by convolutional neural
network (Simonyan and Zisserman, 2014; Kussul et al., 2017; Liu, 2018). Even several mature
convolutional neural network image recognition systems, such as ''Xingse APP'' and ''Aiplants
APP'', have been widely used in the survey of wild plant resources (Gao et al., 2020). However,
the accuracy of these plant image recognition systems is generally low in the classification and
recognition of desert plant images under the conditions of complex nature scenes (Jin, 2020). For
example, Zhang and Huai (2016) used hierarchical deep learning to train and recognize leaf
images of plants with simple scene and complex scenes, and found that the recognition rate of
plants with single scene was as high as 91.11%, while the recognition rate of plants with complex
scenes was only 34.38%. The main problems are as follows. First, the number of images of the
same desert plant species in different nature scenes is too small, and there are fewer images that
can focus on the salient classification characteristics of desert plant species. Second, previous
image recognition system is based on all urban and rural cultivated plants, or a certain organ of
plants, or simple background image datasets. However, in the evolutionary process of long-term
adaptation to the special desert environment, the external morphology of different desert plant
species has similar characteristics (homogenization of plant and branch characteristics, similarity
of branch morphology and color, highly degraded leaf patterns, etc.), which increases the
difficulty of machine learning visual recognition and makes it easier to produce misjudgment (He
et al., 2006). To solve the first problem, researchers proposed the method of obtaining a large
number of plant images conforming to technical requirements (Jin, 2020). The second problem is
the key scientific and technical issue that this study needs to focus on: how to significantly
improve the image recognition accuracy of similar plant species in complex nature scenes and
select the optimal network model suitable for the image recognition of desert plant species, which
is a very challenging task with broad practical application.
1442 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

In view of the lack of research on the image recognition of desert plant species, this study took
the panoramic image set of major desert plant species distributed in nature reserves in Xinjiang
Uygur Autonomous Region of China as the research object, and integrated 37 non-lightweight
and lightweight models of eight categories, such as Visual Geometry Group Network (VGG),
Residual Network (RegNet), and Mobile Network (MobileNet), which are widely used at present
(Krizhevsky et al., 2012; He et al., 2016; Howard et al., 2017). By using grid search to find the
optimal hyperparameters and comparing the performance, we discussed the optimal network
model suitable for the image recognition of desert plant species, so as to achieve convenient and
accurate classification and recognition of desert plant species, and provide a solution for
large-scale field plant background investigation in nature reserves in Xinjiang in the future.

2 Materials and methods

2.1 General survey of nature reserves in Xinjiang
At present, there are 201 protected nature reserves in Xinjiang, which cover an area of 2.51×105
km2, accounting for 15.07% of the total land area of Xinjiang (Fig. 1). Among them, there are one
World Natural Heritage Site, 28 nature reserves, 24 scenic spots, 13 geological parks, 57 forest
parks, 51 wetland parks, and 27 desert parks. In terms of regional distribution, there are 63 in
southern Xinjiang and 138 in northern Xinjiang, accounting for 31.00% and 69.00% of the total
number, respectively.

Fig. 1 Overview of Xinjiang and spatial distribution of nature reserves in Xinjiang. Note that the figure is based
on the standard map (新 S(2021)023) of the Map Service System (https://xinjiang.tianditu.gov.cn/main/bzdt.html)
marked by the Xinjiang Uygur Autonomous Region Platform for Common Geospatial Information Services, and
the standard map has not been modified. Satellite image source: Geospatial Data Cloud (http://www.gscloud.cn/).
2.2 Dataset
Based on the ''List of National Key Protected Wild Plants'' (Ming, 2021), this study selected 24
representative xerophytic desert plant species that are distributed in nature reserves in Xinjiang as
the identification objects (Fig. 2). Since desert plant species were not included in the public
dataset, the images were mainly obtained through field shooting and downloaded from the Plant
Photo Bank of China (PPBC; http://ppbc.iplant.cn/sp/12519). The field collection was extend
from 2019 to 2021. Rangers in nature reserves were commissioned to take pictures with digital
cameras or mobile phones in the natural environment. Those pictures were RGB true color images
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1443

in JPG format. The collected plant images were confirmed by experienced plant experts and
labeled manually. Note that some unclear images were deleted directly. After the sorting process
and statistical analysis, a total of 2331 plant images were finally collected (2071 images from
field collection and 260 images from the PPBC), including 24 plant species belonging to 14
families and 22 genera (Table 1). The training, validation, and test sets were allocated in a ratio of
3:1:1. The plant species information can be found in the Flora of Xinjiang (Xinjiang Flora
Editorial Committee, 1992–2004) and the Red List of Chinese Biodiversity: Higher Plant Volume
(http://www.iplant.cn/rep/protlist/4).

Fig. 2 Images of the selected 24 desert plant species in nature reserves in Xinjiang

2.3 Methods
Convolutional neural network is a branch of deep learning, which is a kind of feedforward neural
network structure containing convolutional computation and with deep structure. In recent years,
it has been widely used in the field of image recognition (Lecun and Bengio, 1998).
Convolutional neural network includes convolutional layer, pooling layer, and fully connected
layer (Fig. 3). The mathematical expression of the network is as follows:
F ( x) = f N ( f N −1 (...( f1 ( x)))), (1)
where x represents the input image; F(x) represents the output of the network, such as the
corresponding class or probability of the input image x; N represents the number of hidden layers;
and fi represents the function of the corresponding layer i.
1444 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

Table 1 Basic information of the selected 24 desert plant species in nature reserves in Xinjiang
Field Total
Species name Family Genera Life form Protection category
images images
Ephedra intermedia Ephedraceae Ephedra Shrub Second-class national 100 114
(00001) protected plant in China

Iljinia regelii Chenopodiaceae Iljinia Subshrub - 50 54

(00002)
Corydalis kashgarica Papaveraceae Corydalis Perennial - 70 94
(00003) herb
Zygophyllum kaschgaricum Zygophyllaceae Zygophyllum Shrub Second-class protected 80 100
(00004) plant in Xinjiang, China
Ammopiptanthus nanus Fabaceae Ammopiptanthus Shrub Second-class national 70 73
(00005) protected plant in China

Oxytropis bogdoschanica Fabaceae Oxytropis Perennial - 90 96

(00006) herb
Caragana polourensis Fabaceae Caragana Shrub - 70 90
(00007)
Glycyrrhiza inflate Fabaceae Glycyrrhiza Perennial Second-class national 50 72
(00008) herb protected plant in China

Ammodendron bifolium Fabaceae Ammodendron Shrub First-class protected plant 15 16

(00009) in Xinjiang, China
Eremosparton songoricum Fabaceae Eremosparton Shrub Second-class protected 30 33
(00010) plant in Xinjiang, China
Lagochilus lanatonodus Lamiaceae Lagochilus Perennial - 30 33
(00011) herb
Frankenia pulverulenta Frankeniaceae Frankenia Annual herb Second-class national 20 32
(00012) protected plant in China

Salsola junatovii (00013) Chenopodiaceae Salsola Subshrub - 105 148

Gymnocarpos przewalskii Caryophyllaceae Gymnocarpos Subshrub First-class national 100 112
(00014) protected plant in China

Helianthemum songaricum Cistaceae Helianthemum. Shrub Second-class national 125 125

(00015) protected plant in China
Haloxylon persicum Chenopodiaceae Haloxylon Tree Second-class national 60 63
(00016) protected plant in China

Caryopteris mongholica Verbenaceae Caryopteris Shrub - 130 149

(00017)
Populus pruinosa Salicaceae Populus Tree First-class protected plant 50 50
(00018) in Xinjiang, China

Tamarix taklamakanensis Tamaricaceae Tamarix Shrub Second-class national 151 151

(00019) protected plant in China

Cistanche deserticola Orobanchaceae Cistanche Perennial Second-class national 100 114

(00020) herb protected plant in China

Calligonum ebinuricum Polygonaceae Calligonum Shrub Second-class protected 105 105

(00021) plant in Xinjiang, China

Prunus tenella Rosaceae Prunus Tree First-class protected plant 20 31

(00022) in Xinjiang, China

Haloxylon ammodendron Chenopodiaceae Haloxylon Tree Second-class national 220 227

(00023) protected plant in China

Populus euphratica Salicaceae Populus Tree - 230 249

(00024)
Note: Values in the parentheses represent the serial numbers and correspond to Figure 1. -, no protection level. Protection category was
referred from the Information System of Chinese Rare and Endangered Plants (ISCREP; https://www.iplant.cn/rep/protlist/3).
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1445

Fig. 3 Schematic diagram of convolutional neural network

In the convolutional layer, f consists of multiple convolution kernels (g1, ..., gk–1, gk), and the
common convolution kernel sizes are 1×1, 3×3, 5×5, and so on. Each gk represents a linear
function in the kth kernel, which can be expressed as follows:
m n w
=g k ( x, y ) ∑ ∑ ∑ Wk (u, v, w) I ( x − u, y − v, z − w), (2)
u=
−m v=
− n w=
−d
where (x, y, z) represents the position of the pixel in the input image I; Wk(u, v, w) represents the
weight of the kernel k; and m, n, and w represent the height, width, and depth of the convolution
kernel, respectively.
In the activation layer, f is a pixel-wise nonlinear function, that is, a rectified linear unit, which
can be represented by the following equation:
f ( x) = max(0, x). (3)
In the pooling layer, f is a layer-wise nonlinear down-sampling function, which aims to
gradually reduce the size of the feature representation.
The fully connected layer can also be considered as a convolutional layer with a kernel size of
1×1. In classification tasks, usually, a prediction layer (i.e., softmax layer) is added to the last
fully connected layer to calculate the probability whether the input images may belong to
different classes. For instance, if the number of neurons in the prediction layer is C (that is, the
number of categories is C): p1, p2, ..., pC, the above C values can be converted to the probability
values through the softmax layer (Eq. 4).
 e pi
pi = C
(i =1, 2, ..., C ), (4)
∑ i =1 e pi
C 
Here, it is set that ∑ i=1 pi = 1 . Finally, the loss function is calculated, and the parameters are
updated through the stochastic gradient descent algorithm. The cross-entropy loss function is one
of the most commonly used loss functions in deep learning, which can measure the difference
between the true value and the predicted value (Li et al., 2020). It is calculated as follows:
C 
loss = −∑ i =1 yi log( pi ), (5)

where yi and pi represent the expected label value and the predicted output value of sample i,
respectively.
In general, the performance of convolutional neural network becomes better as the number of
network layers deepens, such as VGG with 16 layers, Google Inception Network (GoogLeNet)
with 22 layers, and Residual Network (ResNet) with 152 layers (Simonyan and Zisserman, 2014).
However, research shows that no network structure can be guaranteed to outperform other
network structures on any dataset (Liu and Luo, 2019). For a specific dataset, it is necessary to
select the network structure with the best performance according to the experimental results.
Therefore, this study adopts eight categories (including VGG, ResNet, Dense Convolutional
Network (DenseNet), Squeeze Network (SqueezeNet), MobileNet, Shuffle Network (ShuffleNet),
Efficient Network (EfficientNet), and RegNet) of 37 common non-lightweight and lightweight
1446 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

network structures, and adjusts the model parameters to find the best performing network
structure. The experimental environment was: Intel (R)Xeon (R) CPU E5-2640 v3@2.60GHz,
NVIDIA GeForce GTX 2080Ti, Ubuntu 18.04.1. The PyTorch 1.6 deep learning framework was
used, and the batch sizes were set to 4, 8, 16, and 32. Using the Stochastic Gradient Descent
(SGD) optimization algorithm (Li et al., 2021), we determined the following values: a learning
rate size, a momentum of 0.9, and a weight decay rate of 0.005.
2.4 Accuracy evaluation
In this study, the Accuracy, Precision, Recall, and F1 (which refers to the harmonic mean of the
Precision and Recall values) were used to evaluate the model results (Cai, 2020). The Accuracy
measures the ratio of all the correct judgment results of the classification model to the total
samples. Precision is the proportion of the results that are predicted to be positive. Recall refers to
the proportion of all the positive samples that are judged to be positive.
TP + TN
= Accuracy × 100, (6)
TP + FP + TN + FN
TP
Precision
= × 100, (7)
TP + FP
TP
=Recall × 100, (8)
TP + FN
2 × Precision × Recall
= F1 × 100, (9)
Precision + Recall
where TP, TN, FP, and FN represent the numbers of true positive, true negative, false positive,
and false negative samples in the prediction results, respectively.
The complexity of different models is measured by the number of parameters and the number
of floating-point operations (Shen, 2021). The number of parameters refers to the total number of
parameters that need to be trained in the network model, which is used to measure the size of the
model. The number of floating-point operations refers to the number of floating-point operations
per second, which can be used to measure the algorithm complexity. The higher the number of
floating-point operations, the slower the operation speed of convolutional neural network.

3 Results
3.1 Recognition results and comparative analysis of multi models in the image recognition
desert plant species
The model recognition results of plant images are presented in Table 2. Thirteen models with
Accuracy below 70.000% were found, of which the following three were below 55.000%:
SqueezeNet1_0, SqueezeNet1_1, and ShuffleNetV2_X0_5. Twenty-four models with Accuracy
exceeding 70.000% were found, of which the following nine exceeded 75.000%:
EfficientNet_B1, EfficientNet_B3, RegNetX_400MF, RegNetX_800MF, RegNetX_3_2GF,
RegNetX_8GF, RegNetX_16GF, RegNetY_3_2GF, and RegNetY_16GF. RegNetX_8GF
outperformed the other networks, with Accuracy, precision, Recall, and F1 values of 78.333%,
77.654%, 69.547% and 71.256%, respectively.
In addition to the above results, we also compared the number of parameters and the number of
floating-point operations for the different network structures. For the number of parameters, there
were 16 models smaller than 10.000 M (megabyte, which refers to the storage space occupied by
model parameters; 1 M=1024 kilobytes) and four models larger than 100.000 M (VGG11,
VGG13, VGG16, and VGG19). For the number of floating-point operations, there were 15
models smaller than 1.000 G (model operation speed; 1 G=109/s). We used two indicators to
quantify the relationships of the Accuracy with the number of parameters and the number of
floating-point operations (Fig. 4). Amongst the models with an Accuracy higher than 70.000%,
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1447

MobileNetV2 achieves the best balance among the Accuracy, the number of parameters, and the
number of floating-point operations. For MobileNetV2, the Accuracy reaches 71.429%, the
number of parameters is only 2.255 M, the number of floating-point operations is only 0.313 G,
the ratio of Accuracy to the number of parameters is 31.676, and the ratio of Accuracy to the
number of floating-point operations is 228.206. Although RegNetX_8GF exhibited the best
performance, the number of parameters and the number of floating-point operations were 16 and
25 times higher, respectively, compared to MobileNetV2.
Table 2 Experimental results of 37 different models used in the image recognition of desert plant species
Model Accuracy (%) Precision (%) Recall (%) F1 (%) Params (M) FLOPs (G)
VGG11 69.286 68.746 59.769 61.909 128.865 7.613
VGG13 67.857 64.700 58.584 58.310 129.049 11.317
VGG16 65.952 57.170 51.073 51.356 134.359 15.480
VGG19 66.667 67.341 54.269 56.148 139.669 19.643
ResNet18 72.857 75.693 61.621 64.463 11.189 1.819
ResNet34 72.857 74.432 61.448 64.091 21.297 3.671
ResNet50 73.333 69.447 63.473 65.239 23.557 4.110
ResNext50_32_4×D 72.857 71.907 61.531 63.486 23.029 4.257
ResNet101 73.095 66.146 60.646 61.639 42.549 7.832
ResNext101_32_8×D 72.857 73.280 61.886 63.927 86.792 16.475
DenseNet121 65.000 58.949 53.998 54.663 6.978 2.865
DenseNet169 65.476 55.327 54.604 53.783 12.524 3.396
DenseNet201 65.952 59.139 58.259 57.824 18.139 4.339
SqueezeNet1_0 54.286 35.847 34.203 33.751 0.748 0.739
SqueezeNet1_1 53.095 37.520 35.006 33.772 0.735 0.267
MobileNetV2 71.429 64.504 59.919 60.613 2.255 0.313
MobileNetV3-Small 61.429 56.082 51.884 52.685 1.542 0.058
MobileNetV3-Large 64.286 60.755 55.215 56.053 4.233 0.224
ShuffleNetV2_X0_5 53.095 30.013 33.321 30.101 0.366 0.042
ShuffleNetV2_X1_0 60.476 35.104 40.711 36.959 1.278 0.148
EfficientNet_B0 73.810 67.480 62.850 63.810 4.038 0.400
EfficientNet_B1 75.238 75.122 64.517 66.999 6.544 0.591
EfficientNet_B2 74.286 70.714 64.487 66.095 7.735 0.681
EfficientNet_B3 75.714 72.308 65.204 66.359 10.733 0.992
EfficientNet_B4 72.381 68.769 64.278 64.878 17.592 1.543
RegNetX_400MF 76.429 73.074 67.472 68.470 5.105 0.420
RegNetX_800MF 75.000 72.684 65.218 67.185 6.603 0.809
RegNetX_1_6GF 73.333 70.136 64.869 66.038 8.299 1.618
RegNetX_3_2GF 75.000 70.812 64.682 65.630 14.312 3.198
RegNetX_8GF 78.333 77.654 69.547 71.256 37.698 8.021
RegNetX_16GF 75.714 75.210 64.439 66.525 52.279 15.990
RegNetY_400MF 74.762 72.230 65.505 67.071 3.914 0.410
RegNetY_800MF 74.286 70.751 64.156 65.073 5.666 0.845
RegNetY_1_6GF 73.571 70.069 63.407 64.851 10.335 1.629
RegNetY_3_2GF 76.191 70.237 65.338 65.848 17.960 3.200
RegNetY_8GF 74.762 70.841 64.068 64.830 37.413 8.515
RegNetY_16GF 75.000 73.363 66.207 68.029 80.638 15.960
Note: F1 refers to the harmonic mean of the Precision and Recall values. Params, the number of parameters; FLOPs, the number of
floating-point operations. M, megabyte, which refers to the storage space occupied by model parameters (1 M=1024 kilobytes); G,
model operation speed (1 G=109/s). VGG, Visual Geometry Group Network; ResNet, Residual Network; DenseNet, Dense
Convolutional Network; SqueezeNet, Squeeze Network; MobileNet, Mobile Network; ShuffleNet, Shuffle Network; EfficientNet,
Efficient Network; RegNet, Residual Network.
1448 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

Fig. 4 Relationships of the Accuracy with the number of parameters (a) and the number of floating-point
operations (b) for 37 different models used in the image recognition of desert plant species. M, megabyte, which
refers to the storage space occupied by model parameters (1 M=1024 kilobytes); G, model operation speed (1
G=109/s). VGG, Visual Geometry Group Network; ResNet, Residual Network; DenseNet, Dense Convolutional
Network; SqueezeNet, Squeeze Network; MobileNet, Mobile Network; ShuffleNet, Shuffle Network;
EfficientNet, Efficient Network; RegNet, Residual Network.
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1449

3.2 Optimal model for the image recognition of desert plant species
According to the comparative analysis of the above results and considering factors such as
hardware equipment and inference time, MobileNetV2 exhibited the best comprehensive
performance for the image recognition of desert plant species and had better application prospects
in practical work. The classification results of MobileNetV2 and RegNetX_8GF are shown in
Table 3, and the confusion matrix is shown in Figure 5.
Table 3 Classification results of MobileNetV2 and RegNetX_8G in the image recognition of desert plant
species
MobileNetV2 RegNetX_8GF
Species name Precision Recall F1 Precision Recall F1
(%) (%) (%) (%) (%) (%)
Ephedra intermedia (00001) 68.182 68.182 68.182 76.923 90.909 83.333
Iljinia regelii (00002) 60.000 54.546 57.143 66.667 54.546 60.000
Corydalis kashgarica (00003) 75.000 46.154 57.143 100.000 38.462 55.556
Zygophyllum kaschgaricum (00004) 43.750 43.750 43.750 60.000 93.750 73.171
Ammopiptanthus nanus (00005) 90.909 83.333 86.957 73.333 91.667 81.482
Oxytropis bogdoschanica (00006) 58.333 70.000 63.636 50.000 50.000 50.000
Caragana polourensis (00007) 52.632 71.429 60.606 76.923 71.429 74.074
Glycyrrhiza inflata (00008) 80.000 61.539 69.565 80.000 61.539 69.565
Ammodendron bifolium (00009) 0.000 0.000 0.000 0.000 0.000 0.000
Eremosparton songoricum (00010) 100.000 33.333 50.000 100.000 33.333 50.000
Lagochilus lanatonodus (00011) 75.000 50.000 60.000 83.333 83.333 83.333
Frankenia pulverulenta (00012) 0.000 0.000 0.000 100.000 50.000 66.667
Salsola junatovii (00013) 51.220 75.000 60.870 78.261 64.286 70.588
Gymnocarpos przewalskii (00014) 72.727 76.191 74.419 78.261 85.714 81.818
Helianthemum songaricum (000015) 95.833 92.000 93.878 100.000 88.000 93.617
Haloxyon persicum (00016) 25.000 10.000 14.286 50.000 50.000 50.000
Caryopteris mongholica (00017) 82.857 82.857 82.857 85.714 85.714 85.714
Populus pruinosa (00018) 50.000 30.000 37.500 100.000 50.000 66.667
Tamarix taklamakanensis (00019) 73.077 73.077 73.077 73.077 73.077 73.077
Cistanche deserticola (00020) 95.238 90.909 93.023 95.455 95.455 95.455
Calligonum ebinuricum (00021) 75.000 57.143 64.865 80.952 80.952 80.952
Prunus tenella (00022) 75.000 100.000 85.714 100.000 100.000 100.000
Haloxylon ammodendron (00023) 75.000 75.000 75.000 70.175 83.333 76.191
Populus euphratica (00024) 73.333 93.617 82.243 84.615 93.617 88.889

Due to the small amount of data available for Ammodendron bifolium, once the images were
divided into the training, validation, and test sets, the results will be quite different and have no
analytical value. From a Precision perspective for the remaining 23 kinds of plant species, for the
model MobileNetV2, the Precision of all the plant species, except for Oxytropis bogdoschanica
and Haloxylon persicum, was higher than 60.000%. Therefore, MobileNetV2 was able to identify
various types of plant species well and had a high Accuracy. For incorrect classifications, it can
be seen from the confusion matrix that one Caragana polourensis image, one Helianthemum
songaricum image, one Tamarix taklamakanensis image, and two Corydalis kashgarica images
were recognized as Oxytropis bogdoschanica images. Additionally, one Salsola junatovii image,
one Populus pruinose image, one Tamarix taklamakanensis image, one Calligonum ebinuricum
image, and one Haloxylon ammodendron image were recognized as Haloxylon persicum images.
The Recall of Eremosparton songoricum was the lowest, at only 33.000%. The confusion matrix
shows that there were two Eremosparton songoricum images were recognized as Haloxylon
ammodendron in the test set. The Recall of Corydalis kashgarica was the next lowest, at
1450 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

38.500%. Referring to the confusion matrix, it can be seen that in the test set, one image of
Corydalis kaschgarica was respectively recognized as Ammopiptanthus nanus, Lagochilus
lanatonodus, and Haloxylon ammodendron, two images were recognized as Oxytropis
bogdoschanica, and three images were predicted as Caryopteris mongholica.
Upon inspecting the original images of the wrongly classified plant species (Fig. 6), it can be
found that the shape characteristics of desert plants were all presented or the leaves were highly
degraded, which were scaly or cylindrical, due to long-term adaptation to the harsh environment;
or plant shape is approximate to a round spherical. The plant images of Eremosparton songoricum
and Haloxylon persicum in spring and summer are very similar in branch shape, branch color, and
branching pattern. For example, the images taken from the vertical view of Corydalis kashgarica,
Ammopiptanthus nanus, Lagochilus lanatonodus, Haloxylon ammodendron, Oxytropis
bogdoschanica, and Caryopteris mongholica are nearly spherical. Desert plants have different
types of adhesives for stem smoothness or leaf distribution, and the leaves are scaly or cylindrical,
leading to significantly different taxonomic characteristics. In the process of computer vision
recognition, the fine-grained recognition of these fine attributes is not clear, resulting in low
similarity of external morphological features, low image recognition sensitivity, and high false

Fig. 5 Confusion matrix of MobileNetV2 (a) and RegNetX_8GF (b) in the image recognition of desert plant
species. The plant species corresponding to the labels are consistent with those in Figure 2.

Fig. 6 Images of incorrectly classified samples with high similarity of external morphological features
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1451

positive rate of higher plants. This shows that the performance of MobileNetV2 still needs to be
strengthened in recognizing similar but different species of plants, and the image dataset needs to
be improved. From the values of F1 it can be seen that the performances of Eremosparton
songoricum, Corydalis kashgarica, Oxytropis bogdoschanica, and Haloxylon persicum are not
good enough, which are affected by low Precision and Recall values. Considering all the factors,
plants with flowers and fruits or crown shapes and color, such as Populus euphratica, Cistanche
deserticola, Calligonum ebinuricum, Prunus tenella, Lagochilus lanatonodus, and Caryopteris
mongholica, obviously differ from the others without these characteristics in the images and have
better recognition performances (all the indicators exceeding 80.000%). In conclusion, without
the intervention of experts, the lightweight network MobileNetV2 achieves the automatic
classification of plant images accurately and quickly.
3.3 Verifying the research
To verify the validity of the optimal model discovered in this study, we selected the Tianchi
Bogda Peak Nature Reserve and Ebinur Lake Wetland National Nature Reserve, where there are
many desert plants, for empirical verifying. The Ebinur Lake Wetland National Nature Reserve
gathers more than 90.00% of the plant species in the deserts of the Junggar Basin, and some
endangered and endemic species are also distributed there. It is one of the regions with the most
abundant desert plant populations in inland river basins in China, and the plant species here
account for about 64.00% of the country's total desert plant species (Yang et al., 2009). The
Tianchi Bogda Peak Nature Reserve covers the area where the main peak of the eastern Tianshan
Mountains, Bogda Peak, is located. Within a horizontal distance of 80 km from south to north, it
has a complete vertical band spectrum of mountains. With about 700 plant species, the area is the
most typical representative of vertical mountains in the world's temperate arid regions and is
included in the UNESCO Network of Man and Biosphere Programme (Su and Niu, 2016).
Amongst the 24 desert plant species selected in this study, six of them are present in the Ebinur
Lake Wetland National Nature Reserve, and nine of them are distributed in the Tianchi Bogda
Peak Nature Reserve.
The empirical results of MobileNetV2 and RegNetX_8GF in the image recognition of desert
plant species are shown in Table 4. In can be seen that in the image recognition of desert plant
species in the Tianchi Bogda Peak Nature Reserve, the Accuracy, Precision, and Recall of these
models reached 83.000% or more. The Accuracy of MobileNetV2 is 83.871%, which is nearly
5.00% higher than that of RegNetX_8GF in the image recognition of 24 desert plant species
(Accuracy of 78.33%). MobileNetV2 is of high accuracy in the image recognition of desert plant
species, and it has good application prospect in the practical work. In the image recognition of
desert plant species in the Ebinur Lake Wetland National Nature Reserve, each evaluation
indicator also reached more than 60.000% for the two models. It can be seen that both
MobileNetV2 and RegNetX_8GF have high accuracy values. The performance of these two
models was compared in the two nature reserves, with respect to the image recognition of 24
desert plant species. In terms of evaluation indicators, the empirical identification of desert plant
species in the Ebinur Lake Wetland National Nature Reserve and Tianchi Bogda Peak Nature
Reserve was poor, and the values of the evaluation indicators in the Ebinur Lake Wetland
National Nature Reserve were lower than those in the Tianchi Bogda Peak Nature Reserve. There
may be a variety of reasons for this. The images of the nine kinds of desert plant species in the
Tianchi Bogda Peak Nature Reserve were all obtained from the ''Color Atlas of Wild Vascular
Bundle Plants in Bogda Biosphere'' (Su and Niu, 2016), and the pictures were also processed by
screening and clearing. The images of the six kinds of desert plant species in the Ebinur Lake
Wetland National Nature Reserve were taken from the field without clearing and other processing.
These resulted in the differences of the above comparative findings. The comparison also
illustrates the importance of the quality and quantity of image recognition datasets to network
models, and also implies that no one network structure can guarantee its superiority over other
network structures or datasets. For specific datasets, we need to conduct experiments and select
1452 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

the network structure with the best performance, based on the experimental results. This is also
the practical significance of this research.

Table 4 Performances of empirical application of MobileNetV2 and RegNetX_8GF in the image recognition of
desert plant species in the Tianchi Bogda Peak Nature Reserve and Ebinur Lake Wetland National Nature Reserve
Tianchi Bogda Peak Nature Reserve
Model
Accuracy (%) Precision (%) Recall (%) F1 (%)

MobileNetV2 83.871 90.516 83.987 86.715

RegNetX_8GF 86.559 95.299 88.644 91.508

Ebinur Lake Wetland National Nature Reserve

Model
Accuracy (%) Precision (%) Recall (%) F1 (%)
MobileNetV2 64.865 71.466 69.065 68.123
RegNetX_8GF 60.360 77.317 63.309 67.368

4 Discussion
The Accuracy of the optimal MobileNetV2 in the image recognition of desert plant species
screened in this study did not reach more than 90.000% of the image recognition Accuracy of
plant species in a single background, indicating that the finding of this study is still a long way
from practical application (Zhang and Huai, 2016).
Firstly, from the perspective of constructing image dataset, increasing the amount of image data
and enhancing the image quality are helpful to improve the image recognition accuracy (Li,
2022). This study is based on 2331 plant images for model training and testing. At the same time,
due to the large difference in pixel size between plant images collected from the field (using
mobile phones and cameras) and PPBC, the performance of model classifier is affected to some
extent. In the future, through transfer learning, data expansion and image cleaning technology can
solve the problems of insufficient data and different image standards to a certain extent. As for the
problem of complex background in the images, it can be seen from the image recognition results
in the Tianchi Bogda Peak Nature Reserve (Table 4) that the Accuracy can reach more than
80.000% if the object is focused and the features are prominent in an image. Barbedo (2016) also
demonstrated that removing the background of an image can improve the image recognition
Accuracy by 3.000%. However, background removal requires a lot of work and professionals to
complete, which is often difficult to achieve in the application. In theory, the more differential
features extracted from an image, the higher the Accuracy of image recognition (Gai et al., 2021).
It is obvious that leaves, flowers, and fruits of plants have the advantages of multiple shape
features, high recognition, and high discrimination. Future research can fully use the multi-feature
fusion method of panoramic images of plants and images of organs (such as flowers, fruits, and
leaves) to further improve the accuracy and sensitivity of the models.
Secondly, from the perspective of data processing and analysis, fine-grained or new network
structures can be considered to learn and obtain more expressive depth features. It can be seen
from the misclassified plant images (Fig. 6) that due to the similar morphological characteristics
of desert plant species, there is still a problem of high misjudgment rate. On the one hand, the
complexity of the collection environment will cause the uncertainty of expert labeling. However,
Bekker and Goldberger (2016) verified that the deep convolutional network can maintain a high
reliability when the number of mislabeled samples is not very high. On the other hand, the
occurrence of some image recognition errors is probably due to the neglect or inability to
distinguish some subtle attribute features in the process of model learning. For such extreme cases
that cannot be effectively distinguished visually, prior knowledge should be combined to make
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1453

decisions. How to introduce the existing plant family and genus classification labels as prior
information to improve the generalization ability of neural network and make it more suitable for
the image recognition of desert plant species in Xinjiang will become one of the next research
contents (Cao et al., 2018).
Thirdly, in terms of learning algorithm, the current parameter adjustment of convolutional
neural network basically relies on experience and practical operation, which requires constant
training, tuning parameter, and repeated trial and error, consuming a lot of time and energy (Tang,
2020). Auto Machine Learning (AutoML) is the rise of popular research field in recent years (Liu
and Luo, 2019). It automatically builds a network structure, which can guarantee the same
accuracy as classic artificial selection network and its application to the species identification of
natural protected area, and is expected to overcome the artificial selection on subjective fault,
select a better network structure objectively, and improve the image recognition accuracy.
Finally, from the perspective of the scope of application of the models, rare desert plant species
also include Betula holophila, Reaumuria kaschgarica, etc. (Yin, 1991). However, due to the
difficulty of collection, only some desert plant species in Xinjiang were selected as the research
objects in this study. In addition, this study is based on static image data processing and analysis.
At present, a large number of video surveillance systems have been arranged in the nature
reserves in Xinjiang. Therefore, it is also an important direction to strengthen the research on
video image data recognition in the future.

5 Conclusions
Based on image processing and deep learning technology, this study adopted 37 commonly used
non-lightweight and lightweight convolution neural network models of eight categories to
recognize the images of 24 desert plant species typically distributed in Xinjiang. The results show
that there are 24 models with Accuracy above 70.000% and nine models with Accuracy above
75.000%. Among them, the performance of RegNetX_8GF is better than other network models.
The Accuracy, Precision, Recall, and F1 values of RegNetX_8GF are 78.333%, 77.654%,
69.547%, and 71.256%, respectively, which meet the requirements of conventional image
recognition. To further measure the relationships of the Accuracy with the number of parameters
and the number of floating-point operations in the models with Accuracy higher than 70.000%,
we found that MobileNetV2 achieves the best balance among the Accuracy, the number of
parameters, and the number of floating-point operations. The number of parameters for
MobileNetV2 is 1/16 of RegNetX_8GF, and the number of floating-point operations is 1/24.
Considering hardware equipment, inference time, and other factors, MobileNetV2 has the best
performance in the image recognition of desert plant species and is more suitable in field
investigation. In order to verify the effectiveness of this study, we empirically tested
RegNetX_8GF and MobileNetV2 in the image recognition of desert plant species in the Tianchi
Bogda Peak Nature Reserve and the Ebinur Lake Wetland National Nature Reserve, and found
that MobileNetV2 has a good application prospect in the practical work.
Due to the limitations of image datasets, the image recognition accuracy still needs to be
improved. In the future research work, we will further enrich the image sets of desert plant
species in Xinjiang in multiple ways and forms, optimize the convolutional neural network model,
improve the test accuracy, and provide solutions for the administration agencies of nature reserves
to carry out large-scale field plant background investigation, so as to improve work efficiency and
decision-making ability.

Acknowledgements
This work was supported by the West Light Foundation of the Chinese Academy of Sciences (2019-XBQNXZ-
A-007) and the National Natural Science Foundation of China (12071458, 71731009).
1454 JOURNAL OF ARID LAND 2022 Vol. 14 No. 12

References
Abdullahi H S, Sheriff R E, Mahieddine F. 2017. Convolution neural network in precision agriculture for plant image
recognition and classification. In: 2017 Seventh International Conference on Innovative Computing Technology (INTECH).
New York: IEEE, 10: 256–272.
Barbedo J G. 2016. A review on the main challenges in automatic plant disease identification based on visible range images.
Biosystems Engineering, 144: 52–60.
Bargoti S, Underwood J. 2017. Deep fruit detection in orchards. In: 2017 IEEE International Conference on Robotics and
Automation (ICRA). New York: IEEE, doi: 10.48550/arXiv.1610.03677.
Bekker A J, Goldberger J. 2016. Training deep neural-networks based on unreliable labels. In: 2016 IEEE International
Conference on Acoustics, Speech and Signal Processing (ICASSP). New York: IEEE, 2682–2686.
Cai Z. 2020. Research on deep learning model for Chinese herbal medicine planting process. MSc Thesis. Chengdu: University
of Electronic Science and technology. (in Chinese)
Cao X J, Mo Y, Yan Y L. 2020. Convolutional neural network flower image recognition using transfer learning. Computer
Applications and Software, 37(8): 142–148. (in Chinese)
Cao X Y, Sun W M, Zhu Y X, et al. 2018. Plant image recognition based on family priority strategy. Journal of Computer
Applications, 38(11): 3241–3245. (in Chinese)
Coulibaly S, Kamsu F B, Kamissoko D, et al. 2019. Deep neural networks with transfer learning in millet crop images.
Computers in Industry, 108: 115–120.
Gai R L, Cai J R, Wang S Y. 2021. Research review on image recognition based on deep learning. Journal of Chinese Computer
Systems, 42(9): 1980–1984. (in Chinese)
Gao H Y, Gao X H, Feng Q S, et al. 2020. Approach to plant species identification in natural grasslands based on deep learning.
Pratacultural Science, 37(9): 1931–1939. (in Chinese)
Hall D, McCool C, Dayoub F, et al. 2015. Evaluation of features for leaf classification in challenging conditions. In: 2015 IEEE
Winter Conference on Applications of Computer Vision. New York: IEEE, 797–804.
He K M, Zhang X Y, Ren S Q, et al. 2016. Deep residual learning for image recognition. In: 2016 IEEE Conference on
Computer Vision and Pattern Recognition (CVPR). New York: IEEE, 770–778.
He M Z, Zhang J G, Wang H. 2006. Analysis of branching architecture factors of desert plants. Journal of Desert Research, (4):
625–630. (in Chinese)
Howard A G, Zhu M L, Chen B, et al. 2017. Efficient convolutional neural networks for mobile vision applications.
[2022-09-01]. https://arxiv.org/abs/1704.04861v1.
Jin L T. 2020. Research on plant image recognition with complex background based on convolution neural network. MSc
Thesis. Lanzhou: Lanzhou Jiaotong University. (in Chinese)
Krizhevsky A, Sutskever I, Hinton G E. 2012. ImageNet classification with deep convolutional neural networks. Advances In
Neural Information Processing Systems, 25: 1097–1105.
Kussul N, Lavreniuk M, Skakun S, et al. 2017. Deep learning classification of land cover and crop types using remote sensing
data. In: 2017 IEEE Geoscience and Remote Sensing Letters. New York: IEEE, 14(5): 778–782.
Lecun Y, Bengio Y. 1998. Convolutional Networks for Images, Speech, and Time Series. Cambridge, MA: MIT Press, 255–258.
Li L P, Shi F P, Tian W B, et al. 2021. Wild Plant Image recognition method based on residual network and transfer learning.
Radio Engineering, 51(9): 857–863. (in Chinese)
Li M M, Xia W C, Wang M, et al. 2020. Research on monitoring of Chinese nature reserves based on bibliometrics. Journal of
Ecology, 40(6): 2158–2165. (in Chinese)
Li X H, Wu Z H, Liu H, et al. 2020. Species recognition of succulent plants based on convolutional neural network model.
Journal of Guizhou Normal University, 36(3): 9–15. (in Chinese)
Li Y F. 2022. Research on image classification based on optimize on factors in convolutional neural network. Journal of Jinling
Institute of Technology, 38(1): 26–31. (in Chinese)
Liu F Z, Du J H, Zhou Y, et al. 2018. Biodiversity monitoring technology and practice in nature reserves combining UAV and
ground. Biodiversity, 26(8): 905–917. (in Chinese)
Liu H S. 2020. Panoramic plant recognition method based on CNN and GLCM fusion discrimination. MSc Thesis. Wuhan:
Hubei University of Technology. (in Chinese)
Liu Y. 2018. Research on plant recognition based on deep learning. MSc Thesis. Beijing: Beijing Forestry University. (in
Chinese)
Liu Y, Luo Z. 2019. Species recognition of protected area based on AutoML. Computer Systems & Applications, 28(9):
 LI Jicai et al.: Image recognition and empirical application of desert plant species… 1455

147–153. (in Chinese)

Mikolov T, Deo R A, Povey D, et al. 2011. Strategies for training large scale neural network language models. In: 2011 IEEE
Workshop on Automatic Speech Recognition & Understanding. New York: IEEE, 196–201.
Ming Y. 2021. The adjusted "List of National Key Protected Wild Plants" was officially announced. Green China, (19): 74–79.
(in Chinese)
Shen T M J. 2021. Video denoising based on prior information and convolutional neural network. MSc Thesis. Chengdu:
University of Electronic Science and technology. (in Chinese)
Simonyan K, Zisserman A. 2014. Very deep convolutional networks for large-scale image recognition. [2022-09-01].
https://arxiv.org/pdf/1409.1556.pdf.
Su H M, Niu S M. 2016. Color Atlas of Wild Vascular Bundle Plants in Bogda Biosphere. Beijing: China Forestry Press, 10–95.
(in Chinese)
Tang M J. 2020. Research on fast prediction method of ship resistance performance based on convolutional neural network.
MSc Thesis. Harbin: Harbin Engineering University. (in Chinese)
Wang Y W, Tang X L, Xu J P, et al. 2019. The use of big data in nature reserves. China Forestry Economy, (4): 16–20, 27. (in
Chinese)
Xiao Z S. 2019. Application of infrared camera technology in wildlife inventory and assessment of natural reserves in China.
Biodiversity, 27(3): 235–236. (in Chinese)
Xinjiang Flora Editorial Committee. 1992–2004. Xinjiang Flora (Volume I–Volume VI). Urumqi: Xinjiang Science and
Technology Press. (in Chinese)
Yang X D, LV G H, Tian Y H, et al. 2009. Ecological grouping of plants in Lake Abby Wetland Nature Reserve in Xinjiang.
Journal of Ecology, 28(12): 2489–2494. (in Chinese)
Yin L K, Pan B R, Wang Y, et al. 1991. Introduction and cultivation of rare and endangered plants in temperate desert. Arid
Zone Research, (2): 1–8. (in Chinese)
Zhang S, Huai Y J. 2016. Leaf image recognition based on layered convolutions neural network deep learning. Journal of
Beijing Forestry University, 38(9): 108–115. (in Chinese)