We have created Dynamic Gujarati Sign language dataset.Signs are as per ISL-Gujarati.Dataset is captured using webcam. We extracted KeyPoints from each video frame using Media-pipe and stacked them in one variable and save it as .npy file. in eaach npy file there are total 1662 keypoints are there. we have 30 npy files for each sign sample. For each sign we have 625 samples.A total of 36,875 video samples were collected from 10 research participants, both male and female and samples created using data augmentation.
Signs are in this sequence: (['કોઈ સંકેત નથી', 'હા', 'ના', 'તમે', 'હું', 'હફ્તો', 'કાર', 'આભાર', 'બહેરાપણું', 'સારું', 'ખરાબ', 'ખુશ','ઉદાસ','ગુસ્સા','ભાઈ', 'બહેન', 'પરિવાર', 'બાળક', 'છોકરો', 'છોકરી', 'પસંદ', 'ના પસંદ', 'પ્રેમ', 'ઘણું નફોકો', 'સાચું','ખોટું', 'મદદ', 'ઘર', 'કામ', 'શાળા', 'શીખો', 'લાલ', 'વાદળી', 'લીલો', 'સફેદ', 'કાળો', 'દિવસ', 'રાત', 'સાંજ', 'સવાર', 'હવે', 'કાલે', 'ગઈકા','સાઈકલ', 'ટ્રેન', 'પગપાળા', 'દુકાન', 'સસ્તું', 'મહંગું', 'ચા', 'પાણી', 'કોણ','કયો', 'ક્યારે', 'ક્યાં', 'કેમ', 'સરળ', 'મુશ્કેલ', 'સમજાયું'])
Here on the link you will find 24 Pickle files. we have divided whole dataset in 6 parts. Each parts have 4 file. Training and Testing data & Training and Testing labels.