Skip to main content

Table 1 Selection of publicly available facial landmarking databases

From: A review of image-based automatic facial landmark identification techniques

Name Images Subjects Landmarks Description Year
XM2VTS [77] 2 360 295 68ptsa Controlled, 2 head 1999
  (720 × 560)    rotation images, 6 images  
  RGB    captured during speech  
BioID [82] 1 521 23 20 Controlled, front on image, 2001
  (384 × 286)    varying expressions  
  Grayscale     
LFW [84] 13 233 5749 (1680 MULTI-PIE Uncontrolled, originally 2007
  (250 × 250) have ≥ 2 images 68ptsa intended for face identification,  
  mostly RGB in the set)   images collected from the web  
Caltech [81] 7 092 varied size Unknown (10 4pts Uncontrolled, images collected 2007
10 000 mostly RGB 524 faces in   from Google images  
web faces   7 092 images)    
PUT [80] 9 971 100 30 primary Controlled, 2008
  (2048 × 1536) RGB   (194 control portrait images,  
    on subset) 5 poses per subject  
MULTI-PIE [75] 755 370 337 68pts Controlled, 2008
  (3072 × 2048) RGB    landmarks for subset,  
     varying expressions  
MUCT [76] 3 755 276 XM2VTS 68pts +4 Controlled, 5 perspectives 2010
  (640 × 480) RGB   around eyes (76 total) 3 lighting conditions  
     Neutral expression or smile  
AFLW [79] 2 330 varied Unknown PUT 97ptsa Uncontrolled, 2012
  size RGB    portrait images, collected  
     from the web (Flickr)  
HELEN [83] 2 330 Unknown PUT 97ptsa Uncontrolled, collected 2012
  RGB    from the web (Flickr)  
300W [10, 11] 600 varied Unknown MULTI-PIE 68pts Uncontrolled, 300 outdoor 2013
  size RGB    images 300 indoor images  
     difficult poses and expressions  
Menpo benchmark [78] 8 979 varied Unknown Frontal: MULTI-PIE Uncontrolled, 6679 frontal 2017
  size RGB   68 pts Profile: 39pts images 2300 profile images  
     difficult poses and expressions  
  1. aThe Intelligent Behaviour Understanding Group (iBUG) [71] have made MULTI-PIE 68pts landmarks available for this dataset