The pandemic and the politics of Australian research governance

Following the 1989 unified higher education reforms, the Australian academic research system was built upon the notion of depoliticisation (i.e., keeping the political character of decision at one remove from governance) to govern the contradiction between research credibility and governmental economic priorities. The article argues that the COVID-19 pandemic exacerbated the tension between independent research and governmental economic priorities. The pandemic, also, weakened university autonomy via the closure of the national border, reducing overseas student fees, a significant source of research funding. The article maintains that the conservative Morrison government used the opportunity to politicise research around commercialisation and national sovereignty. The argument being that the pandemic exposed Australia's research and development (R&D) dependence and with it the question of industrial sovereignty, prompting the government to couple academic research to industry policy. Secondly, the pandemic reinforced the conservative government's aim to concentrate research in selected commercial areas and to exert this priority on to the research funding agency, the Australian Research Council (ARC). Lastly, the article contends that the COVID pandemic, originating in Wuhan, intensified the Morrison government's geopolitical concerns over China, and this disquiet flowed into research policy, which problematised research collaboration with Chinese researchers.

Plant Stresses and Pest Attack Detection Using Convolutional Neural Networks with Advance Preprocessing Techniques in Chili Plants

Chili is one of the leading commodities worldwide, and in Sri Lanka, it has got high usage because Sri Lankans like spicy foods. Therefore, it is necessary to gain a considerable amount of chilis within a year to export and self-usage. Due to Covid-19, most people are moving to home gardening, and the self-productivity of chilies increased from 2020. However, the problem is, when growing chili plants, several symptoms can occur as deficiencies, diseases, and pest attacks. Early detection and identifying the symptom difference between early-stage and last stage were very hard, and there is no proper accurate application to answer those problems. In modern times, the popularity of transfer learning with Convolutional Neural Network (CNN) is increasing, and there are remaining some paths that not discover yet when detecting plant stresses and pest attacks using the combination of transfer learning and CNN. When considering preprocessing techniques, deblurring techniques usage, object detection, Canny and Sobel edge detection algorithms, and Siamese networks are rarely or never used in this domain. When considering current dataset issues, the variety of illumination conditions of images, lack of use of the real-world dataset, and combination of the real-world dataset with TensorFlow plant village dataset problems are solved in this research. As the research outcome, VGG-19 gives 97.74% top accuracy in deficiency identification using augmentation with deblurring, VGG-16 gives 99.35% top accuracy in pest attack detection using augmentation with Canny filter, and VGG-19 gives 99% top accuracy in disease identification using augmentation with EfficientDet D0 512 × 512. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.