MYLPHerb-1: A Dataset of Malaysian Local Perennial Herbs for the Study of Plant Images Classification under Uncontrolled Environment

Research in the medicinal plants’ recognition field has received great attention due to the need of producing a reliable and accurate system that can recognise medicinal plants under various imaging conditions. Nevertheless, the standard medicinal plant datasets publicly available for research are very limited. This paper proposes a dataset consisting of 34200 images of twelve different high medicinal value local perennial herbs in Malaysia. The images were captured under various imaging conditions, such as different scales, illuminations, and angles. It will enable larger interclass and intraclass variability, creating abundant opportunities for new findings in leaf classification. The complexity of the dataset is investigated through automatic classification using several high-performance deep learning algorithms. The experiment results showed that the dataset creates more opportunities for advanced classification research due to the complexity of the images. The dataset can be accessed through https://www.mylpherbs.com/.

Antisemitism and Covid-19 on Twitter. The search for hatred online between automatisms and qualitative evaluation

The article presents a case study on Antisemitic hate speech in Twitter in the period September 2019 - May 2020, with a particular focus on the months of the Covid-19 emergency. The corpus, consisting of 160.646 tweets selected by keywords, was investigated in terms of the amount of hate for each month, rhetoric and forms of Antisemitism. The analysis is carried out through social network analysis (SNA) techniques, with the goal of understanding whether it is possible to automate the process of identifying Antisemitic hatred. 26.11% of tweets contain hatred, that prejudice is the most common rhetoric (44%) and association with financial power the prevailing form (74%). The sample was also compared with another research methodology that only detects the presence of hate words. It emerges that, in addition to an in-depth knowledge of the phenomenon, it is necessary to integrate the automatic classification phase with the manual contribution. Antisemitismo e Covid-19 in Twitter. La ricerca dell’odio online tra automatismi e valutazione qualitativa. L’articolo presenta un caso studio sul discorso d’odio antisemita in Twitter nel periodo settembre 2019 - maggio 2020, con un particolare affondo sui mesi dell’emergenza Covid-19. Il corpus, composto da 160.646 tweet selezionati per parole chiave, è stato indagato in termini di quantità di odio per mese, retoriche utilizzate e forme di antisemitismo. L’analisi è svolta attraverso le tecniche di social network analysis (SNA), con l’obiettivo di capire se sia possibile automatizzare il processo di individuazione dell’odio antisemita. Il 26.11% dei tweet contiene odio, che il pregiudizio è la retorica più presente (44%) e l’associazione al potere finanziario la forma prevalente (74%). Il campione è stato altresì confrontato con un’altra metodologia di ricerca che rileva la sola presenza di hate words. Emerge che, oltre una conoscenza approfondita del fenomeno, occorre integrare la fase di classificazione automatica con l’apporto manuale.

Comparative Study of Ensemble Models of Deep Convolutional Neural Networks for Crop Pests Classification

Pest infestations on wheat, corn, soybean, and other crops can cause substantial losses to their yield. Early diagnosis and automatic classification of various insect pest categories are of considerable importance for accurate and intelligent pest control. However, given the wide variety of crop pests and the high degree of resemblance between certain pest species, the automatic classification of pests can be very challenging. To improve the classification accuracy on publicly available D0 dataset with 40 classes, this paper compares studies on the use of ensemble models for crop pests classification. First, six basic learning models as Xception, InceptionV3, Vgg16, Vgg19, Resnet50, MobileNetV2 are trained on D0 dataset. Then, three models with the best classification performance are selected. Finally, the ensemble models, i.e, linear ensemble named SAEnsemble and nonlinear ensemble SBPEnsemble, are designed to combine the basic learning models for crop pests classification. The accuracies of SAEnsemble and SBPEnsemble improved by 0.85% and 1.49% respectively compared to basic learning model with the highest accuracy. Comparison of the two proposed ensemble models show that they have different performance under different condition. In terms of performance metrics, SBPEnsemble giving accuracy of classification at 96.18%, is more competitive than SAEnsemble.