publications

2026

1. Preprint

   Sufficient Frame Subsets for Video Understanding

   Lahiru Wijayasingha

   In Under review at WACV 2027, 2026

   Abs

   An explainable AI method for video classification models based on sufficient input subsets: it isolates the smallest set of input frames that drives a model’s decision.

2. AAAI

   EgoEMS: A High-Fidelity Multimodal Egocentric Dataset for Cognitive Assistance in Emergency Medical Services

   Keshara Weerasinghe, Xueren Ge, Tessa Heick, and 5 more authors

   In Proceedings of the AAAI Conference on Artificial Intelligence, 2026

   DOI arXiv Website

2024

1. WACV

   Camera-Independent Single Image Depth Estimation from Defocus Blur

   Lahiru Wijayasingha, Homa Alemzadeh, and John A. Stankovic

   In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024

   Abs HTML PDF Code

   Monocular depth-from-defocus models are sensitive to the camera that captured the images. Using the optics of defocus blur, we show how camera parameters change the blur and propose a correction that makes a trained model work across cameras without retraining, evaluated on a new synthetic dataset, DDFF12 and NYU Depth V2.

2. PRL

   EgoCap and EgoFormer: First-person image captioning with context fusion

   Zhuangzhuang Dai, Vu Tran, Andrew Markham, and 5 more authors

   Pattern Recognition Letters, 2024

   DOI

2022

1. PerCom-W

   Generalized Few-Shot Learning for Wearable Sensor-based Human Activity Recognition

   Lahiru Wijayasingha and John A. Stankovic

   In IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2022

   PDF

2021

1. Smart Health

   Robustness to noise for speech emotion classification using CNNs and attention mechanisms

   Lahiru Wijayasingha and John A. Stankovic

   Smart Health, 2021

   PDF

2016

1. Wireless Health

   A wearable sensing framework for improving personal and oral hygiene for people with developmental disabilities

   Lahiru Wijayasingha and B. P. L. Lo

   In IEEE Wireless Health (WH), 2016

   DOI