Documentation of

Sahand LMC Sign Language Database

Database version 1.0 (2020)
 
 
 
 
Khadijeh Mahdikhanlou, Hossein Ebrahimnezhad
Computer Vision Research Laboratory
Department of Electrical Engineering
Sahand University of Technology
Tabriz, Iran

 

Preface:
This report is the documentation of design, and verification of Sahand LMC Sign Language Database. The database is captured at “Computer Vision Research” Laboratory, Department of Electrical Engineering, Sahand University of Technology. The database is available for academic purposes by filling the access request form.
 
The report is written by:
Khadijeh Mahdikhanlou (kh_mahdikhanlou@sut.ac.ir) &
Hossein Ebrahimnezhad (ebrahimnezhad@sut.ac.ir)
Tel: (+98) 4133459343, Fax: (+98)4133444322
 
 
Sahand LMC Sign Language Database
Sahand LMC Sign Language Database comprises of 32 class including 24 American letters (J and Z are excluded because they are dynamic gestures) and numbers from 0 to 9 (gesture for 6 and w, also gestures for 9 and F are same). Each class of database contains 2000 samples. The database is collected by a webcam and Leap Motion Controller (LMC). The background of the images captured by webcam are relatively simple. The user was asked to move and rotate her hand around the LMC to capture different views of the hand. The database provided by LMC is in structure format. It contains:
     frame id
    timestamp
    pointables
    hands
    version
The pointables class in turn includes these fields:
    id
    position
    velocity
    direction
    is_extended
    is_finger
    is_tool
    is_valid
    length
    width
    touch_distance
    time_visible
And the hands class includes following fields:
    id
    basis
    confidence
    direction
    grab_strength
   is_left
   is_right
   is_valid
   palm_normal
   palm_position
   palm_velocity
   palm_width
   pinch_strength
   sphere_center
   sphere_radius
   stabilized_palm_position
   time_visible
   wrist_position
   arm_elbowPosition
   arm_wristPosition
   arm_direction
   fingers
The fingers structures contains finger_type and bones and each of them include following information:
    basis
    center
    direction
    is_valid
    length
    width
    nextJoint
    prevJoint
    type.
 
Related Work
Khadijeh Mahdikhanlou, Hossein Ebrahimnezhad, "3D American Sign Language Static Alphabet and Numbers Recognition using Hand Joints and Shape Coding", Multimedia Tools and Applications, Springer, Vol., No., pp.
  
 
 
 
 
 
Representation of the database including one sample per each class