gtea dataset - hassony2/inria-research-wiki GitHub Wiki

GTEA dataset project page 2011 [download link]

Egocentric

31k RGB frames

30 fps, 1280x720 pixels (GoPro) 7 types of daily activities, each performed by 4 different subjects

==> 28 videos roughly 1 minute long

(cooking : tea making, sandwiches...)

11 action classes viz., ‘close’, ‘pour’, ‘open’, ‘spread’, ‘scoop’, ‘take’, ‘fold’, ‘shake’, ‘put’, ‘stir’, and ‘background’.

annotation of objects present in sequence (coffee, cup, hotdog, ...) without any temporal and spatial information beyond that.

Provides hand masks (segmentation)