8 October 2018 in IFAC-PapersOnLine
27 March 2025 in Data in Brief
FSFahad Bin Islam Swapnil
Agriculture is a cornerstone of Bangladesh's economy, with tomatoes being one of the most widely cultivated vegetables, producing approximately 368,000 tons annually. However, tomato plants are vulnerable to various diseases and pest infestations that can significantly reduce crop yield, posing a threat to farmers' livelihoods. Early detection of these diseases, often visible through symptoms on the leaves, is critical for effective management. In this work, we present a dataset of 731 high-resolution images of tomato leaves affected by six common diseases, along with healthy samples, aimed at facilitating automated disease diagnosis using computer vision. The dataset is categorized into disease types such as Early Blight, Black Spot, Late Blight, Leaf Mold, Bacterial Spot, and Target Spot. This structured dataset offers a valuable resource for researchers developing machine learning models for disease classification and early detection. By making the dataset publicly available, we aim to accelerate research in precision agriculture and empower the development of AI-driven tools that can enhance tomato disease management, ultimately improving crop yields and supporting sustainable farming practices.
22 December 2023 in AIUB Journal of Science and Engineering (AJSE)
Evaluating teachers' performance is a fundamental pillar of educational enhancement, guiding the evolution of pedagogical practices and fostering enriched learning environments. This study pioneers an innovative approach by harnessing sentiment analysis within an aspect-based framework to decipher the intricate emotional nuances embedded within students' feedback. By categorizing sentiments as positive, negative, and neutral, we delve into the diverse perceptions of teaching aspects, offering a multifaceted portrait of educators' contributions. Through meticulous data collection, preprocessing, and a deep learning sentiment analysis model, we dissected student comments into distinct teaching aspects. The subsequent sentiment analysis unearthed positive, negative, and neutral sentiments. Positive sentiments highlighted strengths and effective communication, while negative sentiments illuminated areas for growth. Neutral sentiments provided contextual equilibrium, forming a holistic tapestry of teachers' performance. The proposed model achieved 86\% F1 score for classifying sentiments into three classes.
13 February 2020 in Scientific Reports
We have investigated the paths taken by Budgerigars while flying in a tunnel. The flight trajectories of nine Budgerigars (Melopsittacus undulatus) were reconstructed in 3D from high speed stereo videography of their flights in an obstacle-free tunnel. Individual birds displayed highly idiosyncratic flight trajectories that were consistent from flight to flight over the course of several months. We then investigated the robustness of each bird’s trajectory by interposing a disk-shaped obstacle in its preferred flight path. We found that each bird continued to fly along its preferred trajectory up to a point very close to the obstacle before veering over the obstacle rapidly, making a minimal deviation to avoid a collision, and subsequently returning to its original path. Thus, Budgerigars show a high propensity to stick to their individual, preferred flight paths even when confronted with a clearly visible obstacle, and do not adopt a substantially different, unobstructed route. The robust preference for idiosyncratic flight paths, and the tendency to pass obstacles by flying above them, provide new insights into the strategies that underpin obstacle avoidance in birds. We believe that this is the first carefully controlled study of the behaviour of birds in response to a newly introduced obstacle in their flight path. The insights from the study could also have implications for conservation efforts to mitigate collisions of birds with man-made obstacles.