Topic of Research Seminar: Deep Learning for Biodiversity Monitoring, Medical Imaging and Multimodal AI Systems

Abstract: The research focuses on the application of deep learning techniques across domains such as biodiversity informatics, biomedical imaging, and multimodal AI. Initial contributions include the use of convolutional neural networks for species identification, trait extraction, and organ-level annotation from digitized herbarium specimens, supporting large-scale ecological data curation. Subsequent work addresses challenges in continual learning, particularly for class-incremental and imbalanced datasets, through data-free generative replay strategies. Recent developments explore synthetic data generation using conditional variational autoencoders (CVAE) for enhancing electromyography-based gesture recognition, and dynamic scene reconstruction leveraging progressive Gaussian splatting and deformation fields. Ongoing and future directions extend into applied AI for environmental and healthcare monitoring. These include UV-based camera traps for automated mosquito surveillance in dengue-prone regions, satellite image analysis for soil texture and moisture estimation, and advanced medical image interpretation using multimodal deep learning frameworks. The research collectively emphasizes scalable, adaptive, and data-efficient AI systems for real-world impact across diverse scientific domains.

Subject field of Topic: Artificial Intelligence, Deep Learning, Computer Vision, Medical Image Analysis, Remote Sensing

Name of Speaker: Dr. Muhammad Sohaib Younis

Professional Rank of Speaker: Assistant Professor

University Email of Speaker: [email protected]

Research Group Weblink: https://smme.nust.edu.pk/faculty/muhammad-sohaib-younis/

Affiliation of Speaker: School of Mechanical and Manufacturing Engineering (SMME-NUST)

Date and Venue: 6th Aug 2025 from 1200hrs to 1300hrs, Seminar Hall, School of Mechanical and Manufacturing Engineering (SMME), NUST Islamabad