Ms. Tazeem Haider recently presented their research work on Human Activity Recognition (HAR) at International Conference on Innovative Computing (ICIC’24) held at UMT Lahore. This prestigious event, published by IEEE, brought together leading experts and researchers to collaborate on innovative computing.

This research addresses growing demands for an efficient, and lightweight activity recognition methods. The study titled “Encoding Human Activity Recognition: Local Coordinate Coding on Multimodal Sensory Data” explores a novel approach for representing and recognizing human activities from raw sensor data.

In her presentation, Tazeem highlighted key limitations of traditional handcrafted techniques and complex deep learning architecture that they lack generalization or are too resource-intensive for deployment on end devices. She pointed out that many systems fail to capture localized and discriminative features effectively, especially when dealing with dynamic motion data from multi modalities. To address these challenges, the study introduces a codebook-based model using Local Coordinate Coding (LCC), which focuses on preserving local geometry in the feature space. This approach enables precise activity recognition, underscores the critical role of efficient representation in wearable sensing and highlights the potential for innovation in low-power activity recognition systems.