[Back]

Closed Eyes In The Wild (CEW)


CEW Dataset [1]

Overview:
Eye closeness detection is a challenging task in the unconstrained real-world application scenario which is full of challenging variations caused by individual difference and kinds of environment changes including lighting, blur, occlusion, and disguise. To investigate the performance of eye closeness detection in these conditions, we collected a dataset for eye closeness detection in the Wild. In particular, this dataset contains 2423 subjects, among which 1192 subjects with both eyes closed are collected directly from Internet, and 1231 subjects with eyes open are selected from the Labeled Face in the Wild (LFW [2]) database. Eye patches are collected based on the coarse face region and eye position automatically and respectively estimated by the face detector[3] and eye localization[4]. We first resize the cropped coarse faces to the size 100100 (pixels) and then extract eye patches of 2424 centered at the localized eye position. Illustration of faces images in this dataset can be seen in Figure 1.

Figure 1: Illustration of the images in CEW dataset. The first row shows faces with both eyes closed, while the second row shows faces with eyes open. The sampled eye patches are shown at the bottom of the figure. Note that these eye patches are full of variances caused by individual, lighting, blur, occlusion, and disguise.

How to Get It?

   We provide three versions of the dataset for you to download: 1) Raw face images with    background, 2) face images warped, and 3) eye patches only, as follows,

(1): Facial images in original resolution, 20M in Rar, download.

 

(2): Facial images in size of 100100, 7.6M in Rar, download.

(3): Eye images in size of 2424, 2.6M in Rar, download.

Other Materials Related to [1]

BioID, CAS-PEAL, AR: Lists of face index with both eyes closed, download

ZJU eyeblink database Eye patches used in [1] from this dataset, download.  

Code:

Histograms of Principal Oriented Gradients (HPOG) and its multi-scale version, 5K in Rar, download.
Visualization of HPOG and HOG descriptors, 33K in Rar, download.

Demo:

Videos:  1   2   3  

References:

[1] F.Song, X.Tan, X.Liu and S.Chen, Eyes Closeness Detection from Still Images with Multi-scale Histograms of Principal Oriented Gradients, Pattern Recognition, 2014.

[2] G. B. Huang, M. Ramesh, T. Berg and E. Learned-Miller, Labeled faces in the wild: A database for studying face recognition in unconstrained environments, Tech. Rep. 07-49, University of Massachusetts, Amherst (October 2007).

[3] P.Viola and M.J.Jones. Robust real-time face detection. International Journal of Computer Vision (IJCV), 57(2):137-154, 2004.

[4] X.Tan, F.Song, Z-H.Zhou and S.Chen. Enhanced Pictorial Structures for Precise Eye Localization under Uncontrolled Conditions, In: Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'09),pp. 1621C1628, Miami, Florida, USA, June 2009.

Copyright and disclaimer:

Copyright 2014, Xiaoyang Tan

The dataset is provided for research purposes to a researcher only and not for any commercial use. Please do not release the data or redistribute this link to anyone else without our permission. Contact {x.tan, f.song}@nuaa.edu.cn if any question.

If you use this dataset, please cite it as,

F.Song, X.Tan, X.Liu and S.Chen, Eyes Closeness Detection from Still Images with Multi-scale Histograms of Principal Oriented Gradients, Pattern Recognition, 2014.