Image features and DUS testing traits for peanut pod variety identification and pedigree analysis.

BACKGROUND: DUS (Distinctness, Uniformity and Stability) testing of new varieties is an important method for peanut germplasm evaluation and identification of varieties. In order to verify the feasibility of variety identification for peanut DUS testing based on image processing, 2000 peanut pod images from 20 varieties were obtained by a scanner. Initially, six DUS testing traits were quantified using a mathematical method based on image processing technology, and then, size, shape, color and texture features (total 31) were also extracted. Next, the Fisher algorithm was used as a feature selection method to select 'good' features from the extracted features to expand the DUS testing traits set. Finally, support vector machine (SVM) and K-means algorithm were respectively used as recognition model and clustering method for variety identification and pedigree clustering. RESULTS: By the Fisher selection method, a number of significant candidate features for DUS testing were selected which can be used in the DUS testing further; using the top half of these features (about 18) ordered by Fisher discrimination ability, the recognition rate of SVM model was found to be more than 90%, which was better than unordered features. In addition, a pedigree clustering tree of 20 peanut varieties was built based on the K-means clustering method, which can be used in deeper studies of the genetic relationship of different varieties. CONCLUSION: This article can provide a novel reference method for future DUS testing, peanut varieties identification and study of peanut pedigree. © 2018 Society of Chemical Industry.

From Organelle Morphology to Whole-Plant Phenotyping: A Phenotypic Detection Method Based on Deep Learning.

The analysis of plant phenotype parameters is closely related to breeding, so plant phenotype research has strong practical significance. This paper used deep learning to classify Arabidopsis thaliana from the macro (plant) to the micro level (organelle). First, the multi-output model identifies Arabidopsis accession lines and regression to predict Arabidopsis's 22-day growth status. The experimental results showed that the model had excellent performance in identifying Arabidopsis lines, and the model's classification accuracy was 99.92%. The model also had good performance in predicting plant growth status, and the regression prediction of the model root mean square error (RMSE) was 1.536. Next, a new dataset was obtained by increasing the time interval of Arabidopsis images, and the model's performance was verified at different time intervals. Finally, the model was applied to classify Arabidopsis organelles to verify the model's generalizability. Research suggested that deep learning will broaden plant phenotype detection methods. Furthermore, this method will facilitate the design and development of a high-throughput information collection platform for plant phenotypes.

Identification of Soybean Mutant Lines Based on Dual-Branch CNN Model Fusion Framework Utilizing Images from Different Organs.

The accurate identification and classification of soybean mutant lines is essential for developing new plant varieties through mutation breeding. However, most existing studies have focused on the classification of soybean varieties. Distinguishing mutant lines solely by their seeds can be challenging due to their high genetic similarities. Therefore, in this paper, we designed a dual-branch convolutional neural network (CNN) composed of two identical single CNNs to fuse the image features of pods and seeds together to solve the soybean mutant line classification problem. Four single CNNs (AlexNet, GoogLeNet, ResNet18, and ResNet50) were used to extract features, and the output features were fused and input into the classifier for classification. The results demonstrate that dual-branch CNNs outperform single CNNs, with the dual-ResNet50 fusion framework achieving a 90.22 ± 0.19% classification rate. We also identified the most similar mutant lines and genetic relationships between certain soybean lines using a clustering tree and t-distributed stochastic neighbor embedding algorithm. Our study represents one of the primary efforts to combine various organs for the identification of soybean mutant lines. The findings of this investigation provide a new path to select potential lines for soybean mutation breeding and signify a meaningful advancement in the propagation of soybean mutant line recognition technology.

Research On Maize Disease Identification Methods In Complex Environments Based On Cascade Networks And Two-Stage Transfer Learning.

Achieving accurate and reliable maize disease identification in complex environments is a huge challenge. This is because disease images obtained from natural environments are often in complex contexts that may contain elements similar to disease characteristics or symptoms. Based on cascade network and two-stage transformation learning, the new method is proposed in this paper and applied the improved method to the task of identification and classification of four maize leaf types in a complex environment. The proposed method has a cascade structure which consists of a Faster R-CNN leaf detector (denoted as LS-RCNN) and a CNN disease classifier, named CENet(Complex Environment Network). The LS-RCNN detector with an attention mechanism was used to detect maize leaves from the image, and the CENet model further classified the leaf images detected in the first stage into four categories: Cercospora leaf spot, Common rust, Northern Leaf Blight, and Healthy, which allowed image features to be extracted more efficiently. The subsequent use of a two-stage transfer learning strategy to train CENet models of disease images in complex contexts allows for faster training of the models while ensuring accuracy. The experimental results show that the proposed method is used to identify four types of maize leaves with an F1-score of 99.70%, which is better than some popular CNN models and others' methods, and has a more obvious advantage in terms of training speed. The model proposed in this experiment has a positive significance for exploring other Crop variety identification and classification under complex backgrounds.

KRS-Net: A Classification Approach Based on Deep Learning for Koi with High Similarity.

As the traditional manual classification method has some shortcomings, including high subjectivity, low efficiency, and high misclassification rate, we studied an approach for classifying koi varieties. The main contributions of this study are twofold: (1) a dataset was established for thirteen kinds of koi; (2) a classification problem with high similarity was designed for underwater animals, and a KRS-Net classification network was constructed based on deep learning, which could solve the problem of low accuracy for some varieties that are highly similar. The test experiment of KRS-Net was carried out on the established dataset, and the results were compared with those of five mainstream classification networks (AlexNet, VGG16, GoogLeNet, ResNet101, and DenseNet201). The experimental results showed that the classification test accuracy of KRS-Net reached 97.90% for koi, which is better than those of the comparison networks. The main advantages of the proposed approach include reduced number of parameters and improved accuracy. This study provides an effective approach for the intelligent classification of koi, and it has guiding significance for the classification of other organisms with high similarity among classes. The proposed approach can be applied to some other tasks, such as screening, breeding, and grade sorting.

Aflatoxin rapid detection based on hyperspectral with 1D-convolution neural network in the pixel level.

Aflatoxin is commonly exists in moldy foods, it is classified as a class one carcinogen by the World Health Organization. In this paper, we used one dimensional convolution neural network (1D-CNN) to classify whether a pixel contains aflatoxin. Firstly we found the best combination of 1D-CNN parameters were epoch = 30, learning rate = 0.00005 and 'relu' for active function, the highest test accuracy reached 96.35% for peanut, 92.11% for maize and 94.64% for mix data. Then we compared 1D-CNN with feature selection and methods in other papers, result shows that neural network has greatly improved the detection efficiency than feature selection. Finally we visualized the classification result of different training 1D-CNN networks. This research provides the core algorithm for the intelligent sorter with aflatoxin detection function, which is of positive significance for grain processing and the prenatal detoxification of foreign trade enterprises.

Pixel-level aflatoxin detecting in maize based on feature selection and hyperspectral imaging.

Aflatoxin is highly toxic and is easily found in maize, a little aflatoxin can induce liver cancer. In this paper, we used hyperspectral data in the pixel-level to build the aflatoxin classifying model, each of the pixel have 600 hyperspectral bands and labeled 'clean' or 'contaminated'. We use 3 method to extracted feature bands, one method is to select 4 hyperspectral bands from other articles: 390 nm, 440 nm, 540 nm and 710 nm, another method is to use feature extraction PCA to obtain first 5 pcs to shrink the hyperspectral volume, the third method is to use Fscnca, Fscmrmr, Relieff and Fishier algorithm to select top 10 feature bands. After feature band selection or extraction, we put the feature bands into Random Forest (RF) and K-nearest neighbor (KNN) to classify whether a pixel is polluted by aflatoxin. The highest accurate for feature selection is Relieff, it reached the accuracy of 99.38% with RF classifier and 98.77% in KNN classifier. PCA feature extraction with RF classifier also reached a high accuracy 93.83%. And the 600 bands without feature extraction reached the accuracy of 100%. Feature bands selected from other papers could reach an accuracy of 89.51%. The result shows that the feature extraction performs well on its own data set. And if the computing time is not taken into account, we could use full band to classify the aflatoxin due to its high accuracy.

Oil Adulteration Identification by Hyperspectral Imaging Using QHM and ICA.

To investigate the feasibility of identification of qualified and adulterated oil product using hyperspectral imaging(HIS) technique, a novel feature set based on quantized histogram matrix (QHM) and feature selection method using improved kernel independent component analysis (iKICA) is proposed for HSI. We use UV and Halogen excitations in this study. Region of interest(ROI) of hyperspectral images of 256 oil samples from four varieties are obtained within the spectral region of 400-720nm. Radiation indexes extracted from each ROI are used as feature vectors. These indexes are individual band radiation index (RI), difference of consecutive spectral band radiation index (DRI), ratio of consecutive spectral band radiation index (RRI) and normalized DRI (NDRI). Another set of features called quantized histogram matrix (QHM) are extracted by applying quantization on the image histogram from these features. Based on these feature sets, improved kernel independent component analysis (iKICA) is used to select significant features. For comparison, algorithms such as plus L reduce R (plusLrR), Fisher, multidimensional scaling (MDS), independent component analysis (ICA), and principle component analysis (PCA) are also used to select the most significant wavelengths or features. Support vector machine (SVM) is used as the classifier. Experimental results show that the proposed methods are able to obtain robust and better classification performance with fewer number of spectral bands and simplify the design of computer vision systems.