An Eye Blink detection technique in video surveillance based on Eye Aspect Ratio

Phuong, Tran Thanh; Hien, Lam Thanh; Toàn, Đỗ Năng; Vinh, Ngo Duc

doi:10.23919/icact53585.2022.9728891

Cited by 10 publications

(5 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• For Blink-AP, InstBlink significantly outperforms the others by large margins (i.e., 21% at least of Blink-AP 50 and 6% at least of Blink-AP 75 ), which verifies the superiority of the proposed framework. We argue that one essential reason is that our framework can model a better long-term temporal eyeblink representation than the frame-based method [9] and the sliding window based methods [19,35,40]. Moreover, under the proposed framework, eyeblink features can be facilitated via face's global context (e.g., head pose and illumination condition) with joint optimization and interaction, while previous works that utilize a sequential manner cannot.…”

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

“…To track each instance, we link the face bounding boxes from the adjacent frames based on their similarity on box IoU. After obtaining the instance tracklets (i.e., a sequence of face bounding boxes and landmarks), we employ 4 representative eyeblink detection method [9,19,35,40] within each instance tracklets. Such a sequential pipeline is commonly used in the existing approaches under the single-person assumption.…”

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

“…Moreover, under the proposed framework, eyeblink features can be facilitated via face's global context (e.g., head pose and illumination condition) with joint optimization and interaction, while previous works that utilize a sequential manner cannot. From Table 2, it can also be summarized that the landmark-based methods [35,40] perform poorer than the region-based counterparts. We think that one essential reason is that landmark detection is unreliable under unconstrained conditions.…”

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

“…Eyeblink detection method. Existing methods generally judge eyeblinks from the pre-extracted local eye clues (e.g., local eye region [6,9,10,19,25] or landmarks around eyes [8,14,35,40]). To obtain the local eye clues, the existing works generally run in a sequential way including face detection, face tracking, and landmark detection.…”

Section: Related Workmentioning

confidence: 99%

See 3 more Smart Citations

Real-time Multi-person Eyeblink Detection in the Wild for Untrimmed Video

Wenzheng¹,

Xiao²,

Wei³

et al. 2023

Preprint

View full text Add to dashboard Cite

Real-time eyeblink detection in the wild can widely serve for fatigue detection, face anti-spoofing, emotion analysis, etc. The existing research efforts generally focus on singleperson cases towards trimmed video. However, multiperson scenario within untrimmed videos is also important for practical applications, which has not been well concerned yet. To address this, we shed light on this research field for the first time with essential contributions on dataset, theory, and practices. In particular, a large-scale dataset termed MPEblink that involves 686 untrimmed videos with 8748 eyeblink events is proposed under multiperson conditions. The samples are captured from unconstrained films to reveal "in the wild" characteristics. Meanwhile, a real-time multi-person eyeblink detection method is also proposed. Being different from the existing counterparts, our proposition runs in a one-stage spatio-temporal way with end-to-end learning capacity. Specifically, it simultaneously addresses the sub-tasks of face detection, face tracking, and human instance-level eyeblink detection. This paradigm holds 2 main advantages: (1) eyeblink features can be facilitated via the face's global context (e.g., head pose and illumination condition) with joint optimization and interaction, and (2) addressing these sub-tasks in parallel instead of sequential manner can save time remarkably to meet the real-time running requirement. Experiments on MPEblink verify the essential challenges of real-time multiperson eyeblink detection in the wild for untrimmed video. Our method also outperforms existing approaches by large margins and with a high inference speed.

show abstract

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

Section: Benchmark Results On Mpeblink Datasetmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Real-time Multi-person Eyeblink Detection in the Wild for Untrimmed Video

Wenzheng¹,

Xiao²,

Wei³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Phuong et al [2] presented a blink detection method based only on eye landmarks. Eye Aspect Ratio (EAR) is calculated from the eye landmarks, and, based on the max and current EAR values, a dynamic threshold is applied for each participant, which determines the eye blink state.…”

Section: Feature-based Methodsmentioning

confidence: 99%

BlinkLinMulT: Transformer-Based Eye Blink Detection

Fodor,

Fenech,

Lőrincz

2023

J. Imaging

View full text Add to dashboard Cite

This work presents BlinkLinMulT, a transformer-based framework for eye blink detection. While most existing approaches rely on frame-wise eye state classification, recent advancements in transformer-based sequence models have not been explored in the blink detection literature. Our approach effectively combines low- and high-level feature sequences with linear complexity cross-modal attention mechanisms and addresses challenges such as lighting changes and a wide range of head poses. Our work is the first to leverage the transformer architecture for blink presence detection and eye state recognition while successfully implementing an efficient fusion of input features. In our experiments, we utilized several publicly available benchmark datasets (CEW, ZJU, MRL Eye, RT-BENE, EyeBlink8, Researcher’s Night, and TalkingFace) to extensively show the state-of-the-art performance and generalization capability of our trained model. We hope the proposed method can serve as a new baseline for further research.

show abstract