Thomas Moeslund's

Gesture Recognition Database


History

The database consists of 2060 images of a hand performing the different static signs used in the international sign language alphabet. The database was constructed in 1996 in connection with a master thesis on gesture recognition, where the gestures to be recognised were the signs. The master thesis is titled "Recognizing Gestures From the Hand Alphabet Using Principal Component Analysis" and can be down loaded here. A condensed version of the work has also been reported in the paper "Real-Time Recognition of Hand Alphabet Gestures Using Principal Component Analysis" which can be downloaded here. Any comments or questions can be emailed to tbm@cvmt.dk

Content

The images in the database are gray-scale images in TIF format with a resolution of 248x256. Each image has been compressed on a Unix machine using the 'gzip' command. Each of the gestures/signs are performed in front of a dark background and the user's arm is covered with a similar black piece of cloth, hence easy segmentation of the hand is possible. Each gesture is performed at various scales, translations, and rotations in the plane parallel to the image-plane. The number of images of each gesture/sign is listed below.

A: 40 B: 60 C: 40 D: 40 E: 40 F: 40 G: 100 H: 100 I: 100 K: 100 L: 100 M: 100 N: 100 O: 100 P: 100 Q: 100 R: 100 S: 100 T: 100 U: 100 V: 100 W: 100 X: 100 Y: 100

The individual images (~37KB) can be down loaded here and the entire data base (tar-file) can be down loaded here (~79MB).


Last modified: Mon Jul 8 14:34:51 2002