Automatic classification of plants based on information content

1970 ◽  
Vol 48 (4) ◽  
pp. 793-802 ◽  
Author(s):  
Laszlo Orloci
Keyword(s):  
Information Theory ◽  
Cluster Analysis ◽  
Information Content ◽  
Random Sample ◽  
Nearest Neighbor ◽  
Theory Model ◽  
Second Stage ◽  
Large Samples ◽  
Two Stages

An information theory model is described and its application is illustrated by an actual example. Classification is accomplished in two stages. The first stage includes cluster analysis of a random sample by an agglomerative method. Cluster analysis is followed by nearest neighbor sorting in the second stage whereby the clustering results are imposed on a second random sample of the same collection. The advantage of the procedure resides in the fact that large samples can be handled, and also, the classification produced in the second stage can be used, under specific restrictive assumptions, for unbiased prediction of different population properties. While the present paper is principally concerned with the technique itself, some taxonomic conclusions are also given.

A novel two-stage method of plant seedlings classification based on deep learning

2021 ◽  
pp. 1-11
Author(s):  
Tianhong Dai ◽  
Shijie Cong ◽  
Jianping Huang ◽  
Yanwen Zhang ◽  
Xinwang Huang ◽  
...  
Keyword(s):  
Deep Learning ◽  
Learning Technology ◽  
Two Stage ◽  
Second Stage ◽  
Stage Classification ◽  
Different Types ◽  
Two Stages ◽  
Plant Seedlings

In agricultural production, weed removal is an important part of crop cultivation, but inevitably, other plants compete with crops for nutrients. Only by identifying and removing weeds can the quality of the harvest be guaranteed. Therefore, the distinction between weeds and crops is particularly important. Recently, deep learning technology has also been applied to the field of botany, and achieved good results. Convolutional neural networks are widely used in deep learning because of their excellent classification effects. The purpose of this article is to find a new method of plant seedling classification. This method includes two stages: image segmentation and image classification. The first stage is to use the improved U-Net to segment the dataset, and the second stage is to use six classification networks to classify the seedlings of the segmented dataset. The dataset used for the experiment contained 12 different types of plants, namely, 3 crops and 9 weeds. The model was evaluated by the multi-class statistical analysis of accuracy, recall, precision, and F1-score. The results show that the two-stage classification method combining the improved U-Net segmentation network and the classification network was more conducive to the classification of plant seedlings, and the classification accuracy reaches 97.7%.

Classification of cultivated barley (Hordeum vulgare). 1. Historical aspects, and phenetic character analysis of some characters by information theory and by spatial autocorrelation

1986 ◽  
Vol 64 (11) ◽  
pp. 2769-2773
Author(s):  
Bernard B. Baum
Keyword(s):  
Information Theory ◽  
Hordeum Vulgare ◽  
Spatial Autocorrelation ◽  
Theory Model ◽  
Kernel Weight ◽  
Hordeum Vulgare L ◽  
Department Of Agriculture ◽  
Character Analysis ◽  
Barley Cultivars

A brief historical sketch of the classification of barley (Hordeum vulgare L.) cultivars is presented along with reference to key reviews on this subject. Characters, utilized in the comprehensive study on the barley cultivars of North America by Aberg and Wiebe (U.S. Department of Agriculture Technical Bulletin 942), were subjected to a series of phenetic character analyses using an information theory model and a spatial autocorrelation model. The ranking of the 48 characters in order of their importance (for classification and identification purposes) from the character analysis by information theory was compared with the previous rating of characters made by Aberg and Wiebe and was found to differ significantly. Numerous trials of character analysis by spatial autocorrelation using various Minkowski distances, setting various values among three parameters, never yielded results comparable with those obtained by Aberg and Wiebe. Among those trials, a few combinations of values for the three parameters (X, Y, and Z) yielded results comparable with those obtained with character analysis by information theory. Those same combinations of values were found by Estabrook and Gates (Taxon, 33: 13–25) in their study of Banisteriopsis in 1984, where they also developed the method of character analysis by spatial autocorrelation. Kernel weight was found to be the most important character.

scholarly journals Predictions in avalanche forecasting

2000 ◽  
Vol 31 ◽  
pp. 377-381 ◽  
Author(s):  
D. M. McClung
Keyword(s):  
Field Data ◽  
Informational Entropy ◽  
Avalanche Forecasting ◽  
The Public ◽  
Second Stage ◽  
Relevant Field ◽  
Verification Of Models ◽  
Using Data ◽  
Two Stages

AbstractVerification of avalanche forecasts depends on the spatial and temporal scale of the forecast, and the classes of informational entropy of data implicit in the forecast. First I present a classification system for avalanche forecasts based on these parameters. Verification of models in avalanche forecasting may consist of two stages. Often, the first stage is to ensure that the model matches the scales (space and time) and the classification of forecast and that redundant variables and parameters are eliminated. Once that is achieved, verification can proceed to the second stage, testing the model against relevant field data and situations. I provide an example based on the public-danger scale bulletin used for warnings in the back country in North America and Europe. Using data on deaths and accidents from Alpine Europe with Bayesian statistics, I conclude the danger scale has more classes than necessary for back-country applications. This could be a first stage prior to actual verification of this experience-based model.

scholarly journals Natural Language Processing Based Instrument for Classification of Free Text Medical Records

2016 ◽  
Vol 2016 ◽  
pp. 1-10 ◽  
Author(s):  
Manana Khachidze ◽  
Magda Tsintsadze ◽  
Maia Archuadze
Keyword(s):  
Language Processing ◽  
Medical Records ◽  
Nearest Neighbor ◽  
Health Management ◽  
Support Vector ◽  
Free Text ◽  
Features Selection ◽  
Second Stage ◽  
History Of

According to the Ministry of Labor, Health and Social Affairs of Georgia a new health management system has to be introduced in the nearest future. In this context arises the problem of structuring and classifying documents containing all the history of medical services provided. The present work introduces the instrument for classification of medical records based on the Georgian language. It is the first attempt of such classification of the Georgian language based medical records. On the whole 24.855 examination records have been studied. The documents were classified into three main groups (ultrasonography, endoscopy, and X-ray) and 13 subgroups using two well-known methods: Support Vector Machine (SVM) andK-Nearest Neighbor (KNN). The results obtained demonstrated that both machine learning methods performed successfully, with a little supremacy of SVM. In the process of classification a “shrink” method, based on features selection, was introduced and applied. At the first stage of classification the results of the “shrink” case were better; however, on the second stage of classification into subclasses 23% of all documents could not be linked to only one definite individual subclass (liver or binary system) due to common features characterizing these subclasses. The overall results of the study were successful.

Hierarchical Classification Of Expert Models Of Exercises Designed To Eliminate Specific Mistakes Occurring In Short Ski Turn

2020 ◽  
Author(s):  
Danijela Kuna ◽  
Matej Babić ◽  
Mateja Očić
Keyword(s):  
Hierarchical Classification ◽  
Vertical Position ◽  
Chi Square ◽  
Second Stage ◽  
Expert Model ◽  
Significant Difference ◽  
Hands On ◽  
Chi Square Test ◽  
Two Stages

The aim of the present study was to examine the structure of an expert model of exercises designed to eliminate the Lack of specific ski movement mistake in short ski turn, as well as offer a hierarchical classification of the expert model. For this purpose, a two-stage research was conducted. During the first stage of the research the exercises with the purpose of Lack of specific ski movement mistake elimination were designed by 20 skiing experts aged 25 to 45. By means of email and coordinated by the paper author, the experts first designed a model of 14 methodical exercises and subsequently selected the five most relevant ones, ranking them on a scale from 1 to 5. A nonparametric chi - square test (χ2) was used. The research showed there was no significant variation across the experts’eval-uation of the five most important methodical exercises (χ2 = 21,69; p = 0,06). The expert model of the most important methodical exercises for the Lack of specific ski movement mistake correction thus includes the following: Holding a ski stick under the handle, Jump turns, Hands on hips, Unbuttoned ski boots and Ski poles in vertical position in forwards. 307 skiing professionals of various levels of expertise participated in the second stage of the research, whose aim was to classify the Lack of specific ski movement mistake elimi-nation exercises. The participants’task was to rank the exercises based on their relevance. Total amounts of rank sums (ΣR) were calculated, the Kruskal-Wallis test (H-test) was car-ried out, and the corresponding levels of significance (p) were recorded, for the purpose of comparing the significance of diversity between rank sums and the expert model. The sta-tistically significant difference was found between the rank sums (ΣR) of the most eficient exercises for the Lack of specific ski movement mistake correction (H = 198,19; p < 0,001). The results obtained in the two stages of the research provide valuable insights regarding the methods of short ski turns. The hierarchical classification of the most important method-ical corrective exercises obtained from ski teachers and professionals with different levels of education and expertise yields accurate and precise data about corrective methodical exercises in the process of studying short ski turn. Any further research regarding the same object should evaluate the designed expert model of the most important methodical exer-cises, as well as their hierarchical classification, across different groups of participants.

scholarly journals Classification of Pap smear images for Cervical cancer using Convolutional Neural Network

2019 ◽  
Vol 9 (1) ◽  
pp. 2801-2807
Keyword(s):  
Neural Network ◽  
Cervical Cancer ◽  
Convolutional Neural Network ◽  
Pap Smear ◽  
Severe Dysplasia ◽  
Normal Cells ◽  
Second Stage ◽  
Two Stages ◽  
Cancerous Cells

Classification of Pap smear images for cervical cancer consists of two types namely, normal and abnormal cancerous cells. The dataset involves 7 sets of classes of cancerous images which have 3 sets of normal cancerous images and 4 sets of abnormal cancerous images. The proposed work performs two stages of classification. The first stage of the work is classifying the data as normal or abnormal cancerous cells. In the second stage of the work, the class of the cancer as normal columnar, normal intermediate, normal superficial, light dysplasia, moderate dysplasia, severe dysplasia and carcinoma_in_situ are classified. The proposed work gives good results for classifying images for 3 sets of classes and 4 sets of classes for normal cells and is able to classify and detect normal and abnormal cell accurately.

Classification of vegetation data from an open beach environment in southwestern Ontario: cluster analysis followed by generalized distance assignment

1975 ◽  
Vol 53 (5) ◽  
pp. 495-502 ◽  
Author(s):  
Gary E. Bradfield ◽  
László Orlóci
Keyword(s):  
Cluster Analysis ◽  
Plant Communities ◽  
Orthogonal Transformation ◽  
Similarity Matrix ◽  
Topographic Features ◽  
Two Stages ◽  
Type Classification ◽  
Generalized Distance ◽  
Beach Environment

Vegetation data from a tract of open beach in southwestern Ontario were classified in two stages to aid in the description and mapping of the plant communities of the area. Firstly, the similarity matrix generated for half the sample was analyzed by a method of similarity clustering designed to produce homogeneous and distinct groups. The four groups that emerged conformed with the main topographic features in the study area, these being the shoreline zone, the middle beach zone, the wet slack zone, and the upper beach zone. This four-type classification was then imposed upon the rest of the sample, using generalized distance (D2) as the assignment function. The problem faced in the inversion of the singular group covariance matrix (5k) for each group was overcome by orthogonal transformation. Although considerable computation was involved, the results indicated that D2, when used in a deterministic sense, has much potential in helping to allocate individuals to groups.

Implementation of a Hybrid Classification Method for Diabetes

2019 ◽  
pp. 201-240
Author(s):  
Dilip Kumar Choubey ◽  
Sanchita Paul ◽  
Kanchan Bala ◽  
Manish Kumar ◽  
Uday Pratap Singh
Keyword(s):  
Feature Selection ◽  
Decision Tree ◽  
Fuzzy Decision ◽  
Fuzzy Decision Tree ◽  
Feature Selection Technique ◽  
Second Stage ◽  
Classification Technique ◽  
Hybrid Classification ◽  
Two Stages

This chapter presents a best classification of diabetes. The proposed approach work consists in two stages. In the first stage the Pima Indian diabetes dataset is obtained from the UCI repository of machine learning databases. In the second stage, the authors have performed the classification technique by using fuzzy decision tree on Pima Indian diabetes dataset. Then they applied PSO_SVM as a feature selection technique followed by the classification technique by using fuzzy decision tree on Pima Indian diabetes dataset. In this chapter, the optimization of SVM using PSO reduces the number of attributes, and hence, applying fuzzy decision tree improves the accuracy of detecting diabetes. The hybrid combinatorial method of feature selection and classification needs to be done so that the system applied is used for the classification of diabetes.

SEGMENTATION OF LUNG PATTERNS IN HIGH-RESOLUTION COMPUTED TOMOGRAPHY IMAGES OF THE LUNG

2008 ◽  
Vol 07 (03) ◽  
pp. 265-280 ◽  
Author(s):  
ALENA SHAMSHEYEVA ◽  
ARCOT SOWMYA ◽  
PETER WILSON
Keyword(s):  
Region Of Interest ◽  
Ground Glass Opacity ◽  
High Resolution Computed Tomography ◽  
Second Stage ◽  
Knowledge Based ◽  
Automated Method ◽  
Computed Tomography Images ◽  
High Level ◽  
Two Stages

An automated method is presented for segmentation of two-dimensional HRCT images of the lung into regions of four lung patterns: normal, emphysema, honeycombing, and ground-glass opacity (GGO). Segmentation was implemented in two stages. At the first stage, pixel-wise classification of the lung area was performed using local textural features extracted by the wavelet transform. At the second stage, classification results were refined by application of knowledge-based rules. Performance of the method was compared on two sets of HRCT images: one included HRCT images with characteristic examples of lung patterns and the other consisted of unselected HRCT images that represented a model of routine operations at a general radiology practice. On the first set of images, sensitivity of the method ranged from 0.92 to 0.99, and specificity ranged from 0.96 to 0.99. On the second set of images, sensitivity and specificity were, respectively, 0.49 and 0.95 for emphysema, 0.87 and 0.55 for normal, 0.34 and 0.99 for honeycombing, and 0.57 and 0.94 for GGO. The two-stage approach allowed for simple and effective application of high-level knowledge about appearance of lung patterns on HRCT images and did not require feature and region of interest size selection for the first stage of pixel-wise lung pattern classification.

Algorithm of adaptive correction of structural landforms on global digital elevation models

2018 ◽  
Vol 937 (7) ◽  
pp. 57-64
Author(s):  
A.Y. Zhdanov ◽  
A.V. Pankin ◽  
A.V. Rentel
Keyword(s):  
High Altitude ◽  
Digital Elevation Models ◽  
Reference Network ◽  
Height Difference ◽  
Second Stage ◽  
Digital Elevation ◽  
Adaptive Correction ◽  
Dem Accuracy ◽  
Two Stages

Due to various factors, such as the interpolation step or automatic correlators specifics, global digital elevation models (DEM) often have an effect of understating the heights, which leads to inaccurate display of structural landforms e.g. ridges. The algorithm of adaptive correction of structural landforms elevation on DEM is proposed in this article. The algorithm consists of two stages. In the first stage, an automatic classification of structural forms is performed based on height difference between neighboring DEM elements. In the second stage, the DEM elements are corrected based on the assigned classes. Adaptivity of the algorithm allows to use it for any kind of terrain and elevation ranges. The algorithm was tested on the global DEM ALOS World 3D (ALOS W3D30); the accuracy was assessed by geodetic reference network and ICESat mission data. The developed algorithm shows an improvement of DEM accuracy, especially in high-altitude areas, and it also helps to reveal areas requiring additional verification.

Sign in / Sign up

Export Citation Format

Share Document