Dicta-Sign Training Data for Isolated Signs

Dicta-Sign Training Data for Isolated Signs

The video data found on this page is mainly meant as machine learning resources:

Icon for Training Data

DGS data with little variation from two different signers: Annotated videos of 5+5 copies of each sign in isolation

Each token is marked with exact start and end position of the sign. The signs produced correspond to the 1000 concepts list.

Sample video with available tagging as subtitles

Session name	Video files					Annotation files		Metadata
	A frontal view	A over head view	B frontal view	B over head view	overview	ELAN	iLex	IMDI
dicta_sign_voc_01
dicta_sign_voc_02
dicta_sign_voc_03
dicta_sign_voc_04
dicta_sign_voc_05

Icon for Training Data

DGS data with some variation from two different signers: Annotated videos of about 5+5 copies of each sign in isolation

Only the time interval in which all copies of a sign are performed is marked. The signs produced correspond to the 1000 concepts list.

Sample video with available tagging as subtitles

Session name	Video files					Annotation files		Metadata
	A frontal view	A over head view	B frontal view	B over head view	overview	ELAN	iLex	IMDI
dicta_sign_voc_sept_01
dicta_sign_voc_sept_01a
dicta_sign_voc_sept_02
dicta_sign_voc_sept_02a
dicta_sign_voc_sept_03
dicta_sign_voc_sept_03a
dicta_sign_voc_sept_04
dicta_sign_voc_sept_05

Icon for Training Data

GSL data with some variation from one signer: Annotated videos of about 5 copies of each sign in isolation

The signs produced correspond to the 1000 concepts list.