Leaf Image Dataset Research Articles

Leaf Image Recognition Based on Bag of Features

Plants are ubiquitous in human life. Recognizing an unknown plant by its leaf image quickly is a very interesting and challenging research. With the development of image processing and pattern recognition, plant recognition based on image processing has become possible. Bag of features (BOF) is one of the most powerful models for classification, which has been used for many projects and studies. Dual-output pulse-coupled neural network (DPCNN) has shown a good ability for texture features in image processing such as image segmentation. In this paper, a method based on BOF and DPCNN (BOF_DP) is proposed for leaf classification. BOF_DP achieved satisfactory results in many leaf image datasets. As it is hard to get a satisfactory effect on the large dataset by a single feature, a method (BOF_SC) improved from bag of contour fragments is used for shape feature extraction. BOF_DP and LDA (linear discriminant analysis) algorithms are, respectively, employed for textual feature extraction and reducing the feature dimensionality. Finally, both features are used for classification by a linear support vector machine (SVM), and the proposed method obtained higher accuracy on several typical leaf datasets than existing methods.

Plant species recognition based on global–local maximum margin discriminant projection

Plant species recognition using leaves is an important and challenging research topic, because the plant leaves are various and irregular and they have very large within-class difference and between-class similarity. Considering that leaves have different discriminant performance and contribution to plant recognition task, based on maximum neighborhood margin discriminant projection (MNMDP), we propose a global–local maximum margin discriminant projection (GLMMDP) algorithm for plant recognition. GLMMDP utilizes the local and class information and the global structure of the data to model the intra-class and inter-class neighborhood scatters and a global scatter, obtaining the projection matrix by minimizing the local intra-class scatter and meanwhile maximizing both the local inter-class scatter and the global between-class scatter. Compared with MNMDP, GLMMDP not only can detect the true intrinsic manifold structure of the data, but also can enhance the pattern discrimination between different classes by incorporating the global between-class scatter into MNMDP. The global between-class scatter fully indicates the difference and similarity between classes. The experimental results on the ICL (Intelligent Computing Laboratory) leaf datasets and Leafsnap leaf image datasets demonstrate the effectiveness of the proposed plant recognition method. The recognition accuracy is more than 95% on the ICL datasets and more than 90% on Leafsnap datasets.

Learned features of leaf phenotype to monitor maize water status in the fields

Water stress significantly influences normal maize growth. Fast and effective maize water stress detection is of great help to monitor the plant status and provide scientific guidance for crop irrigation. Most of the methods are based on manual measurements of soil water content, or laboratory imaging techniques, such as hyperspectral and thermal images at plant level. With the collection of 656 original maize plant images under natural environment, a novel maize leaf image dataset with different water stress levels (well-watered, reduced-watered and drought-stressed) was constructed. This paper considers maize water status detection as a fine-grained classification problem using local leaf images. Inspired by deep learning, a convolutional neural network (CNN) is applied for the first time to maize water stress recognition. In the designed CNN architecture, feature maps from different convolutional layers are merged. Through visualization and importance analysis of the multi-scale feature maps, several specific feature maps are selected as learned features, which provide a strong discrimination ability. An SVM classifier is finally trained using the feature representation as inputs. Compared with existing techniques, the proposed method achieves the satisfying classification performance with an accuracy of 88.41%. This study also provides a quantitative measure of water stress degree using a regression model. Experimental results demonstrate that the learned features perform better than hand-crafted features to detect water stress and quantify stress severity. The proposed framework can be deployed in practical applications for a non-destructive, near real-time, and automatic monitoring of plant water status in fields.

Mathematical and Visual Understanding of a Deep Learning Model Towards m-Agriculture for Disease Diagnosis

Disease detection and classification based on the disease spot found on the leaves is of great importance in improving the agricultural productivity. This paper provides a comprehensive overview of the prevailing applications of computer vision and deep learning techniques in the field of agriculture highlighting the necessity of disease identification and classification using leaf image dataset. A novel classification framework is proposed explaining its working principle. The proposed framework is applied on the multispace image reconstruction inputs. The multispace image reconstruction inputs are used to generate a new set of images containing its gradient images. Then high level semantic features are extracted from the original and reconstructed images, via. convolutional and depthwise separable convolutional layers. Finally, softmax classifier is used for classification. The hyperparameters and computational cost are computed mathematically which provides an insight of creativeness to the researchers. The framework performance is evaluated and compared with the related works on publicly available apple leaf image dataset.

Research on deep learning in apple leaf disease recognition

The main reason affecting apple production is the occurrence of apple leaf diseases, which causes huge economic losses every year. Therefore, it is of great significance to study the identification of apple leaf diseases. Based on DenseNet-121 deep convolution network, three methods of regression, multi-label classification and focus loss function were proposed to identify apple leaf diseases. In this paper, the apple leaf image data set, including 2462 images of six apple leaf diseases, were used for data modeling and method evaluation. The proposed methods achieved 93.51%, 93.31% and 93.71% accuracy on the test set respectively, which were better than the traditional multi-classification method based on cross-entropy loss function with an accuracy of 92.29%.

Neural network–based leaf classification using machine learning

SummaryIn this paper, the leaf image data set is taken for classification. The process is done as image pre‐processing, image segmentation, feature extraction, and image classification. The segmentation is done on colours. The K‐Means clustering is applied to group the similar colour pixels. In total, three sub‐images are created as an output of segmentation. In total, three classes are considered for the process. Data set with 70 leaf images is considered for the classification process. The data set are classified as yellow‐based leaves, brown‐based leaves, and green‐dominated leaves. In this, the first two classes are considered as infected, based on the colours, and the third class is an uninfected collection of leaves. Seven different Neural Networks are constructed for classification process using MATLAB. Their performances are evaluated using the confusion matrices. From the outcome of confusion matrix, it is clear that the Regression Neural Network and Radial Bias Neural Network are better classifiers out of the seven architectures.

Integrating leaf and flower by local discriminant CCA for plant species recognition

Plant species recognition using a single organ, such as flower and leaf, is not sufficiently reliable, because different species may have very similar flowers or leaves, while the same species may have rather different flowers or leaves. Combining leaves and flowers to recognize plant species can produce positive results. Based on multi-modal learning scheme, an automatic plant species recognition method is proposed by combining leaves and flowers of plant. In the method, a modified local discriminant canonical correlation analysis (MLDCCA) is designed by incorporating the idea of local discriminant embedding (LDE) into canonical correlation analysis (CCA). Firstly, two neighbor graphs are constructed based on the exploration of the manifold that the input data lie on. Then, two projection matrices for dimensionality reduction are obtained by making the within-class neighbor samples most correlated and between-class neighbor samples least correlated, and meanwhile keeping the correlation between leaves and flowers of the same species maximum. Finally, 1-nearest neighbor classifier with geodesic distance is used to recognize the plant species. MLDCCA is a powerful supervised multi-modal dimensional reduction method which can extract the discriminant features from two plant organs, meanwhile preserve the discriminant information and the data structure well. Experimental results on a real leaf and flower image dataset validate the effectiveness of the proposed method.

Research on leaf species identification based on principal component and linear discriminant analysis

A method of leaf species identification based on principal component and linear discriminant analysis is proposed in this paper. The method consists of four phases. Leaf images obtained by cameras or other devices typically have petioles and wormholes that affect the recognition rate. Hence, it is necessary to first preprocess the images before feature extraction. After preprocessing of the leaf samples, the binary image, grayscale image, and texture image of the leaves are output for feature extraction. Then, shape features and texture features are extracted to describe the leaves. The shape features can be divided into three groups: geometric characteristics, Hu moment invariants features, and structural characteristics. The texture features consists of gray level co-occurrence matrix features, fractal dimension features, Local Binary Patterns features, and Gabor features. Next, principal component analysis and linear discriminant analysis are combined to reduce the dimension of the features. Finally, Back Propagation Neural Network is used to classify the feature data. The proposed method was tested on two leaf image datasets: Flavia and ICL; the average accuracy of leaf species identification was 94.22 and 87.82%, respectively. The experiment demonstrated the effectiveness of the proposed method.

Leaf image based cucumber disease recognition using sparse representation classification

Most existing image-based crop disease recognition algorithms rely on extracting various kinds of features from leaf images of diseased plants. They have a common limitation as the features selected for discriminating leaf images are usually treated as equally important in the classification process. We propose a novel cucumber disease recognition approach which consists of three pipelined procedures: segmenting diseased leaf images by K-means clustering, extracting shape and color features from lesion information, and classifying diseased leaf images using sparse representation (SR). A major advantage of this approach is that the classification in the SR space is able to effectively reduce the computation cost and improve the recognition performance. We perform a comparison with four other feature extraction based methods using a leaf image dataset on cucumber diseases. The proposed approach is shown to be effective in recognizing seven major cucumber diseases with an overall recognition rate of 85.7%, higher than those of the other methods.

An Automatic Leaf Disease Detection System for Legume Species

Legumes are crucial species which are used by the community worldwide. In this manuscript, a two stage approach to identify infected leaf region percentage in legumes (particularly Groundnut and Soybean) is proposed. First stage classifies between a healthy and a diseased leaf sample. Second stage detects the diseased region and identifies the proportion of leaf infected area. The two stage approach provides high accuracy and also, shows that texture features plays an important role for classification of healthy and diseased leaves. The experimental results obtained on a self-collected leaf image dataset show that the proposed approach accurately identifies the diseased region in legumes. The proposed methodology can also be used for the classification of different disease types.

Similarity based leaf image retrieval using multiscale R-angle description

Leaf image identification is a significant and challenging application of computer vision and image processing. A central issue associated with this task is how to effectively and efficiently describe the leaf images and measure their similarities. In this paper, a novel shape descriptor termed R-angle is proposed. R-angle describes the curvature of the contour by measuring the angle between the intersections of the shape contour with a circle of radius R centered at points sampled around the contour. It is intrinsically invariant to group transforms including scaling, rotation and translation. Varying the parameter R of the proposed R-angle naturally introduces the notation of scale, which we leverage to provide a coarse-to-fine description of the local curvature. A local scale arrangement is proposed by taking the distance between each contour point and the center of the shape to be the maximum scale for a given contour point. Two matching schemes, including L1-norm matching and dynamic programming based matching, are applied to measure the similarities of the leaf shapes. The retrieval experiments conducted on two challenging leaf image datasets indicate that the proposed method significantly outperforms the state-of-the-art methods for leaf identification. An additional experiment on an animal dataset also indicates its potential for general shape recognition.

A Novel Method of Automatic Plant Species Identification Using Sparse Representation of Leaf Tooth Features.

Automatic species identification has many advantages over traditional species identification. Currently, most plant automatic identification methods focus on the features of leaf shape, venation and texture, which are promising for the identification of some plant species. However, leaf tooth, a feature commonly used in traditional species identification, is ignored. In this paper, a novel automatic species identification method using sparse representation of leaf tooth features is proposed. In this method, image corners are detected first, and the abnormal image corner is removed by the PauTa criteria. Next, the top and bottom leaf tooth edges are discriminated to effectively correspond to the extracted image corners; then, four leaf tooth features (Leaf-num, Leaf-rate, Leaf-sharpness and Leaf-obliqueness) are extracted and concatenated into a feature vector. Finally, a sparse representation-based classifier is used to identify a plant species sample. Tests on a real-world leaf image dataset show that our proposed method is feasible for species identification.

PNN based crop disease recognition with leaf image features and meteorological data

An automatic crop disease recognition method was proposed in this paper, which combined the statistical features of leaf images and meteorological data. The images of infected crop leaves were taken under different environments of the growth periods, temperature and humidity. The methods of image morphological operation, contour extraction and region growing algorithm were adopted for leaf image enhancement and spot image segmentation. From each image of infected crop leaf, the statistical features of color, texture and shape were extracted by image processing, and the optimal meteorological features with the highest accuracy rate were obtained and selected by the attribute reduction algorithm. The fusion feature vector of the image was formed by combining the statistical features and the meteorological features. Then the probabilistic neural networks (PNNs) classifier was adopted to evaluate the classification accuracy. The experimental results on three cucumber diseased leaf image datasets, i.e., downy mildew, blight and anthracnose, showed that the crop diseases can be effectively recognized by the integrated application of leaf image processing technology, the disease meteorological data and PNNs classifier, and the recognition accuracy rate was higher than 90%, which indicated that the PNNs classifier trained on the disease feature coefficients extracted from the crop disease leaves and meteorological data could achieve higher classification accuracy. Keywords: image processing, crop disease recognition, disease meteorological data, morphology, probabilistic neural networks (PNNs) DOI: 10.3965/j.ijabe.20150804.1719 Citation: Shi Y, Wang X F, Zhang S W, Zhang C L. PNN based crop disease recognition with leaf image features and meteorological data. Int J Agric & Biol Eng, 2015; 8(4): 60－68.

Plant identification using leaf shapes—A pattern counting approach

Plant identification is required by all walks of life, from professionals to the general public. Nevertheless, it is not an easy job but requires specialized knowledge. In this paper, we propose a new method for plant identification using shapes of their leaves. Different from existing studies which target at simple leaves, the proposed method can accurately recognize both simple and compound leaves. In specifics, we propose a novel feature that captures global and local shape information independently so that they can be examined individually during classification. Furthermore, we advocate that when comparing two leaf individuals it is better to “count” the number of certain shape patterns rather than to match the extracted shape features in a point-wise manner. The proposed counting-based shape descriptor is not only discriminative for classification but also computationally fast and storage cheap. Experiments conducted on five leaf image datasets demonstrate that our algorithm significantly outperforms the state-of-the-art methods in terms of recognition accuracy, efficiency and storage requirement.

MARCH: Multiscale-arch-height description for mobile retrieval of leaf images

In this paper, we propose a novel shape description method for mobile retrieval of leaf images. In this method, termed multiscale arch height (MARCH), hierarchical arch height features at different chord spans are extracted from each contour point to provide a compact, multiscale shape descriptor. Both the global and detailed features of the leaf shape can be effectively captured by the proposed algorithm. MARCH descriptors are compared using a simple L1-norm based dissimilarity measurement providing very fast shape matching. The algorithm has been tested on four publicly available leaf image datasets including the Swedish leaf dataset, the Flavia leaf dataset, the ICL leaf dataset and the scanned subset of the ImageCLEF leaf dataset. The experiments indicate that the proposed method can achieve a higher classification rate and retrieval accuracy than the state-of-the-art benchmarks with a more than 500 times faster retrieval speed. A mobile retrieval system embedding the proposed algorithms has been developed for the real application of leaf image retrieval.

Active subclustering

Although there are many excellent clustering algorithms, effective clustering remains very challenging for large datasets that contain many classes. Image clustering presents further problems because automatically computed image distances are often noisy. We address these challenges in two ways. First, we propose a new algorithm to cluster a subset of the images only (we call this subclustering), which will produce a few examples from each class. Subclustering will produce smaller but purer clusters. Then we make use of human input in an active subclustering algorithm to further improve results. We run experiments on a face image dataset and a leaf image dataset and show that our proposed algorithms perform better than baseline methods.

Citrus disease recognition based on weighted scalable vocabulary tree

Citrus Huanglongbing (HLB) is a destructive disease in citrus production that causes huge economic damage to citrus producers and related industries in the world. Early and accurate detection of HLB is a critical management step to control the spread of this disease. However, existing HLB detection methods cannot be widely adopted in citrus production due to long-time and high-cost detection period in specific laboratory environments. In view of this, a fast-response and low-cost computer vision technique is investigated for diagnosing HLB in citrus leaves. Specifically, the Gaussian mixture density (GMD) is performed to extract the leaf object from the citrus image, followed by the feature extraction and recognition of the existence of HLB in the leaf based on scalable vocabulary tree. A citrus leaf image dataset is constructed, and the experimental results show that the proposed HLB recognition method with GMD object extraction performs 95–100 % accuracy within 1 s.

Comparative Study of Feature Extraction Components from Several Wavelet Transformations for Ornamental Plants

Human has a duty to preserve the nature, preserving the plant is one of the examples. This research emphasis on ornamental plant that has functionality not only as ornament plant but also as a medicinal plant. Purpose of this research is to find the best of the particular feature extraction components from several wavelet transformations. It consists of Daubechies, Dyadic, and Dual-tree complex wavelet transformation. Dyadic and Dual-tree complex wavelet transformations have shift invariant property. While Daubechies is a standard wavelet transform that widely used for many applications. This comparison is utilizing leaf image datasets from ornamental plants. From the experiments, obtained that best classification performance attained by Dual-tree complex wavelet transformation with 96.66% of overall performance result.