Blink2Speak: An Eye Blink and Gaze-Based Human Computer Interface for Paralysed Patients

This literature review examined recent advancements in Human Computer Interface (HCI) systems developed for paralysed and physically challenged individuals. The review focused on understanding how modern vision-based, machine learning, and AI-assisted technologies improve communica- tion, cursor control, accessibility, and independence for users who are unable to operate traditional input devices such as keyboards and mice. The reviewed studies demonstrate significant progress in eye tracking, blink detection, gaze estimation, gesture recognition, and intelligent assistive communication systems. Recent re- search works based on OpenCV, Convolutional Neural Net- works (CNN), Hidden Markov Models (HMM), adaptive gaze calibration, and semi-supervised learning frameworks have considerably improved real-time cursor control accuracy and usability [1], [2], [3], [4], [5], [6]. These systems provide more reliable interaction when compared with earlier traditional image-processing approaches. The literature also highlights the growing importance of gesture-based and touchless interaction systems for physi- cally challenged individuals [7], [8], [9], [10]. Systems using hand gestures, head movement detection, and virtual keyboard communication provide multiple communication pathways and improve user accessibility without depending on expensive hardware devices. Another major observation from the reviewed studies is the increasing adoption of low-cost webcam-based assistive technologies. Several systems successfully utilize standard webcams and open-source frameworks such as OpenCV, Me- diaPipe, and Dlib for gaze tracking and blink detection [2], [5], [7]. This significantly reduces implementation cost and improves the feasibility of practical deployment in hospitals, rehabilitation centers, and home environments. Despite these advancements, several limitations still exist. Many systems continue to face challenges related to lighting sensitivity, false blink detection, cursor instability, calibration dependency, latency, and reduced performance under dynamic real-world conditions [16], [17], [18]. In addition, prolonged usage may lead to user fatigue, especially in systems relying heavily on continuous eye gaze or head movement interaction [11], [12], [13]. The literature review also identifies important research gaps related to adaptive learning, personalized interaction, predictive text generation, and large-scale real-time validation [19], [20], [21]. Future research should therefore focus on developing more intelligent, adaptive, and user-centric assis- tive communication systems capable of operating efficiently in practical environments. Overall, the reviewed studies strongly support the devel- opment of an affordable, accurate, and real-time eye blink communication system for paralysed users. The findings pro- vide a strong theoretical and technological foundation for the proposed Blink2Speak system and justify the need for further research in intelligent assistive Human Computer Interface technologies [22], [23], [24].

[1]     S. Heo and R. Manduchi, “Mouse-Guided Gaze: Semi-Supervised Learning of Intention-Aware Representations for Reading Detection,” NeurIPS 2025 Workshop / arXiv preprint arXiv:2509.19574, 2025. [2]     P. Narsimha, P. Spandana, Mounika, Manjula, and G. Spandana, “Eye Ball Cursor Movement Using Open CV,” Journal of Science Engineering Technology and Management Science, vol. 2, no. 8, pp. 326–332, 2025. [3]     C. P. V et al., “AI-Driven Neural Eye Cursor for Paralyzed Patients,” SENNET, 2025. [4]     G. R. Chhimpa et al., “A Comprehensive Framework for Eye Tracking,” Journal of Eye Movement Research, 2025. [5]     G. Tanusha et al., “Enabling Cursor Control through Eye Movement using Hidden Markov Model,” 2025. [6]     D. Abdullah and S. H. Nazeer, “An Intelligent Eye Gaze Virtual Keyboard with Real Time Blink Interaction,” 2025. [7]     C. Liu et al., “Enhancing Smartphone Eye Tracking,” CHI, 2025. [8]     A. H. Khaleel et al., “Multitask Virtual Keyboard Controlled by an Eye- Gaze-Based Interface,” 2025. [9]     Y. K. Meena and M. Salvi, “Multimodal Appearance Based Gaze- Controlled Virtual Keyboard,” 2025. [10]    “Vision-Based Methods for Virtual Mouse and Keyboard Interaction,” 2025.