BananaLSD: A banana leaf images dataset for classification of banana leaf diseases using machine learning.

Bananas, one of the most widely consumed fruits globally, are highly susceptible to various leaf spot diseases, leading to significant economic losses in banana production. In this article, we present the Banana Leaf Spot Diseases (BananaLSD) dataset, an extensive collection of images showcasing three prevalent diseases affecting banana leaves: Sigatoka, Cordana, and Pestalotiopsis. The dataset was used to develop the BananaSqueezeNet model [1]. The BananaLSD dataset contains 937 images of banana leaves collected from banana fields, which were then further augmented to generate another 1600 images. The images were acquired using three smartphone cameras in diverse real-world conditions. The dataset has potential for reuse in the development of machine learning models that can help farmers identify symptoms early. It can be useful for researchers working on leaf spot diseases and serve as motivation for further researches.

COVIDXception-Net: A Bayesian Optimization-Based Deep Learning Approach to Diagnose COVID-19 from X-Ray Images.

COVID-19 is spreading around the world like wildfire. Chest X-rays are used as one of the primary tools for diagnosing COVID-19. However, about two-thirds of the world population do not have access to sufficient radiological services. In this work, we propose a deep learning-driven automated system, COVIDXception-Net, for diagnosing COVID-19 from chest X-rays. A primary challenge in any data-driven COVID-19 detection is the scarcity of COVID-19 data, which heavily deteriorates a deep learning model's performance. To address this issue, we incorporate a weighted-loss function that ensures the COVID-19 cases are given more importance during the training process. We also propose using Bayesian Optimization to find the best architecture for detecting COVID-19. Extensive experimentation on four publicly available COVID-19 datasets shows that our proposed model achieves an accuracy of 0.94, precision 0.95, recall 0.94, specificity 0.997, F1-score 0.94, and Matthews correlation coefficient 0.992 outperforming three widely used architectures-VGG16, MobileNetV2, and InceptionV3. It also surpasses the performance of several state-of-the-art COVID-19 detection methods. We also performed two ablation studies that show our model's accuracy degrades from 0.994 to 0.950 when a random search is used and to 0.983 when a regular loss function is employed instead of the Bayesian and weighted loss, respectively.