A low rank model based improved eye detection under spectacles

Human video database for facial feature detection under spectacles with varying alertness levels: a baseline study

“…Datasets like [6, 9, 10, 16, 23, 25] allow the subjects to enact low‐alertness condition. Such en‐action results in a false analysis of the relationship between the alertness level and the visual cues. Accessories (here: spectacles): Articles in [2, 4, 7, 23, 45] suggest that eye feature detection and thus the alertness level is vulnerable to the use of spectacles by the subject. Though [10, 25–28] have created databases of facial images with spectacles, the variation in the spectacle frames introduced by them is limited.…”

“…Accessories (here: spectacles): Articles in [2, 4, 7, 23, 45] suggest that eye feature detection and thus the alertness level is vulnerable to the use of spectacles by the subject. Though [10, 25–28] have created databases of facial images with spectacles, the variation in the spectacle frames introduced by them is limited.…”

“…Spectacles commonly used facial accessories critically impact ocular feature detection and tracking [7, 45, 46, 48, 56]. Occlusion, glare, and specular reflections, as shown in Fig.…”

“…A more generalised de‐occlusion algorithm is reported in Zhao et al [59]. Lazarus et al [45] are based on the assumption that facial images subjected to various illumination conditions tend to lie on a low‐dimensional Lambertian space. Therefore, a given sequence of multiple images is decomposed into the Low rank and Sparse component.…”