Orthy Toor | Searcharxiv

arXiv2026

Binary Contact Sensing for Sitting Posture Recognition Without Pressure Sensors

Prolonged sitting with poor posture results in musculoskeletal injury. Early intervention and prevention methods to monitor posture rely on cameras, wearable devices, and dense pressure arrays. Although effective, these approaches introduce privacy concerns, calibration needs, higher cost, etc. In this paper, we explore the spatial pattern of body contact as a binary posture feature vector for a distinct contact-based sitting posture recognition system without the need for analog signal conditioning and calibration. Our sensing principle uses 10 mechanical contact switches arranged in a 5 x 2 array on the backrest. Each switch encodes local body contact into 10-bit binary posture signatures for four postures: normal sitting, leaning back, leaning left, and leaning right. Decision tree and logistic regression classifiers achieved highest accuracy of 96%. Low correlation amongst the switches indicates that every switch captures complimentary information for successful posture classification. SHAP analysis identified the central contacts as the most informative and significant region for posture discrimination. Our results show that binary contact patterns can capture sufficient spatial information for reliable posture recognition. This sensing strategy offers a simple & low-cost alternative to pressure-based systems to monitor posture without the need for mapping the biomechanical pressure distribution.

eess.SP↗

arXiv2026

Single-Cycle Multidirectional EOG Classification Faster than Human Reaction Time for Wearable Human-Computer Interactions

Electrooculogram (EOG) is a non-invasive bio-signal generated by the potential difference between the retina and cornea during eye movement, and is widely utilized in Human-Computer Interaction (HCI) systems. Expanding the range of detectable eye movements enhances system capability. However, increasing the number of classes typically degrades classification performance. While AI-based approaches can mitigate this limitation, their complexity increases significantly when operating on single-cycle EOG signals. Although single-cycle signals offer advantages such as low latency, reduced power consumption, and improved responsiveness, they are inherently limited by reduced informational content and higher susceptibility to noise. Ensuring low latency remains critical for real-time HCI applications, where system response must remain below human reaction thresholds. In this experimental study, using explainable AI, we address these challenges by developing 1-dimensional (1D) and cascaded ANN and CNN architectures capable of highly accurate classification across ten EOG classes (Stare, Blink, Up, Down, Right, Left, Up-left, Up-right, Down-left, and Down-right) using single-cycle signals, while simultaneously achieving latency substantially lower than human reaction time. The study achieved an accuracy of around 99% for all the models with a latency of 38.6 ms for the 1D ANN, and 82.85 ms for the cascaded CNN. These findings confirm that cascaded neural network architectures, can effectively balance high classification accuracy and low latency for single-cycle, multi-class EOG-based HCI systems under limited data availability.

eess.SP↗