Advancements in computer vision, particularly in image segmentation and object detection, are pivotal for marine ecosystem monitoring. This research investigates real-time object detection using masked convolutional neural networks implemented via Detectron2, leveraging the Google Open Images v7 dataset to evaluate precision, recall, Intersection over Union (IoU), and mean Average Precision (mAP). Through multiple training iterations under a transfer-learning paradigm, the study showcases the effectiveness of the proposed framework for automated, scalable marine ecosystem monitoring.

This paper compares computing platforms suitable for deep learning applications, then focuses on predicting the torque required by an exoskeleton system designed to assist users in lifting objects with minimal effort. Using sensor data embedded within the exoskeleton, torque predictions are performed in an offline, real-time setting. The results demonstrate the effectiveness of predictive modelling for accurately estimating torque requirements from historical sensor data, with broader implications for assistive technologies.

Despite profound transformation in the automotive industry, traditional vehicle diagnostics still present significant challenges for owners and mechanics alike. This study develops an Intelligent Conversational Assistant for automobile fault diagnostics, built with Android Studio and Kotlin, and demonstrates the potential of AI and natural language processing for better diagnostic accuracy and improved user experience.