Conference
SenseLess: Minimal Vision, Maximum Insight for Smart Homes
IEEE International Conference on Pervasive Computing and Communications (PerCom), 2026. (To appear)
Research Programme
Augmenting Anomaly Detection with Tiny Cameras
(2021-2025)
Computer Vision
Edge Computing
Low-Cost Cameras
Anomaly Detection
Image Processing
Internet of Things (IoT)
Infrastructure / Systems (IS)
Security (S)
Human Computer Interaction (HCI)
Project Overview
Smart home environments rely on sensor networks to detect unusual patterns and behaviours, yet traditional sensors such as temperature, humidity, and motion detectors often miss contextual cues that a visual observation would readily reveal. This project investigates how miniature, low-cost cameras can augment existing anomaly detection systems, adding a layer of visual intelligence without compromising occupant privacy or affordability. The research builds prototypes that combine simple image capture with edge analytics to enrich anomaly detection pipelines. By processing images locally on resource-constrained devices, the approach minimises data transfer to external servers and preserves occupant privacy by design. The project systematically studies trade-offs between camera placement strategies, lighting conditions, image resolution, and the fusion of visual signals with other sensor modalities to improve detection accuracy in residential IoT settings.
Understanding how camera-based and traditional sensor signals interact is essential for designing practical visual augmentation systems. Through experimental evaluation in smart home testbeds, the work assesses how different levels of visual information contribute to identifying anomalies that purely numerical sensor data would overlook. These evaluations cover diverse environmental conditions and device configurations to ensure generalisable findings.
The findings inform practical deployment guidelines for integrating low-cost computer vision into existing smart environment infrastructures. By establishing clear recommendations on resolution, placement, and sensor fusion strategies, the project helps practitioners adopt visual augmentation while maintaining acceptable privacy guarantees for building occupants.
Team
Norah Abdullah Albazzai
Yaser Abu Awwad
Charith Perera