SWP-LeafNET: A novel multistage approach for plant leaf identification based on deep CNN

Beikmohammadi, Ali; Faez, Karim; Motallebi, Ali

doi:10.1016/j.eswa.2022.117470

Cited by 23 publications

(6 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the Flavia dataset, as shown in Table 9, the proposed algorithm achieves a classification accuracy of 99.3%, which is in the third place and little lower than the accuracy of 99.65% obtained by Wu et al 38 combining shape IMTD and convolutional features (VGG16 + ReLU5_2) and 99.67% obtained by Ali et al 39 modeling a botanist's behavior in leaf identification by proposing a highly efficient method of maximum behavioral resemblance developed through three deep learning-based models (SWP-LeafNET). The Pearline's 32 and Ganguly's 51 methods based on deep learning performs slightly poor and obtains a classification accuracy of 98.7%, lower than the proposed algorithm.…”

Section: Overall Evaluation Of the Plant Classification Systemmentioning

confidence: 71%

“…The recognition results on nine benchmark leaf datasets for the unsupervised leaf image retrieval and supervised leaf image classification are better than prior state-of-the-art plant identification approaches. Beikmohammadi et al 39 proposed a highly efficient method of maximum behavioral resemblance developed through three deep learning-based models to model a botanist's behavior, in which the first and second models are designed from scratch, and the third model employed the pre-trained architecture MobileNetV2 along with the transfer-learning technique. S-LeafNET, W-LeafNET, and P-LeafNET is distributable and considerably faster than other methods because of using shallower models with fewer parameters asynchronously.…”

Section: Contour-based Shape Featuresmentioning

confidence: 99%

See 1 more Smart Citation

Effective shape features for leaf classification

Zhang,

Dai,

Long

et al. 2023

J. Electron. Imag.

View full text Add to dashboard Cite

.When conducting image classification, traditional and deep-learning based methods require extracting target features in different ways and then classifying targets based on the features. Therefore, extracting more informative and effective features is very important, and this problem also exists in plant leaf classification. To address the issue for traditional leaf classification, this article carefully designs two shape features [area ratio-based shape descriptor (ARSD) and corner-based shape descriptor (CSD)] and one leaf vein feature (LVF) extraction method based on statistical characteristics. Then, according to the extraction of ARSD, CSD, and LVF, a general feature extraction framework is formed, which first normalizes direction, segments region of interest, then divides it into subregions, counts the number of pixels in each sub region, and finally forms feature vectors. Finally, to visually verify the effectiveness of these features, they were applied to the classic leaf classification framework, which is adjusted to form a two-stage classification method from coarse classification of shapes and fine classification of species. In the first stage (coarse classification), leaves in three universal datasets [Flavia dataset, Swedish dataset and Intelligent Computing Laboratory (ICL) dataset] and a self-built LZU dataset are respectively divided into several shape categories only by the ARSD. Experiment results show that the ARSD has obvious advantages in shape-based rough classification. Then in the second stage (fine classification), shape, texture, vein, and color features, including ARSD, CSD, local binary patterns, gray-level co-occurrence matrix, the developed LVF, and color moments, are combined to subdivide finer species in each shape category. Extensive experiment results show that the classification accuracy on the Flavia, Swedish, ICL, and LZU datasets have reached 99.3%, 98.9%, 91.5%, and 95.3%, respectively, comparable to deep learning-based methods.

show abstract

Section: Overall Evaluation Of the Plant Classification Systemmentioning

confidence: 71%

Section: Contour-based Shape Featuresmentioning

confidence: 99%

Effective shape features for leaf classification

Zhang,

Dai,

Long

et al. 2023

J. Electron. Imag.

View full text Add to dashboard Cite

show abstract

“…Finally, to build an reliable system, the authors of [57] have proposed an efficient method of behavioral similarity developed through three models based on deep learning. To train their models, they have used the MalayaKew dataset that includes 44 classes of plant species and the FLavia dataset that contains 32 plant species.…”

Section: Capturementioning

confidence: 99%

Artificial Intelligence for Automated Plant Species Identification: A Review

Labrighli¹,

Moujahdi²,

Oualidi³

et al. 2022

IJACSA

View full text Add to dashboard Cite

Plants are very important for life on Earth. There is a wide variety of plant species and their number increases each year. The plants identification using conventional keys is complex, takes time and it is frustrating for non-experts because of the use of specific botanical terms/techniques. This creates a difficult obstacle to overcome for novices interested in acquiring knowledge about species, which is very important to develop any environmental study, like climate change anticipation models for example. Today, there is an increasing interest in automating the species identification process. The availability and omnipresence of relevant technologies, such as digital cameras, mobile devices, pattern recognition and artificial intelligence techniques in general, have allowed the idea of automated species identification to become a reality. In this paper, we present a review of automated plant identification over all significant available studies in literature. The main result of this synthesis is that the performance of advanced deep learning models, despite the presence of several challenges, is becoming close to the most advanced human expertise.

show abstract

“…However, this manual identification process is labor-intensive and prone to subjective misunderstandings, introducing a risk of misinterpretation [7]. In view of this, numerous researchers have proposed intelligent recognition methods for plant diseases, and they have used CNN models to classify plant images with good recognition results [8][9][10][11][12][13]. These plant images are categorized based on their background into two types: natural background images and pure background images.…”

Section: Introductionmentioning

confidence: 99%

AMS-MLP: Adaptive Multi-Scale MLP Network with Multi-Scale Context Relation Decoder for Pepper Leaf Segmentation

Fang,

Jiang,

Liu

et al. 2024

Preprint

View full text Add to dashboard Cite

Pepper leaf segmentation plays a crucial role in monitoring pepper leaf diseases in various backgrounds and ensuring the healthy growth of peppers. However, existing transformer-based segmentation methods suffer from computational inefficiency, excessive parameterization, and limited utilization of edge information. To tackle these challenges, we propose an adaptive multi-scale MLP framework, named AMS-MLP, which combines the multi-path aggregation module (MPAM) and the multi-scale context relation mask module (MCRD) to refine the object boundaries in pepper leaf segmentation. AMS-MLP consists of an encoder-based network, an adaptive multi-scale MLP (AM-MLP) module, and a decoder network. In the encoder network, the MPAM module effectively fuses five-scale features to generate a single-channel mask, improving the accuracy of pepper leaf boundary extraction. The AM-MLP module divides the input features into two branches: the global multi-scale MLP branch captures long-range dependencies between image information, while the local multi-scale MLP branch focuses on extracting local feature maps. Adaptive attention mechanism is designed to dynamically adjust the weights of global and local features. The decoder network incorporates the MCRD module into the convolutional layer, enhancing the extraction of boundary features. To verify the performance of the proposed method, we conducted extensive experiments on three pepper leaf datasets with different backgrounds. The results demonstrate mIoU scores of 97.39%, 96.91%, and 97.91%, as well as F1 scores of 98.29%, 97.86%, and 98.51%, respectively. Comparative analysis with U-Net and state-of-the-art models reveals that the proposed method dramatically improves the accuracy and efficiency of pepper leaf image segmentation.

show abstract

SWP-LeafNET: A novel multistage approach for plant leaf identification based on deep CNN

Cited by 23 publications

References 33 publications

Effective shape features for leaf classification

Effective shape features for leaf classification

Artificial Intelligence for Automated Plant Species Identification: A Review

AMS-MLP: Adaptive Multi-Scale MLP Network with Multi-Scale Context Relation Decoder for Pepper Leaf Segmentation

Contact Info

Product

Resources

About