GeorgiaTech Egocentric Activities (GTEA)


This dataset contains 7 types of daily activities, each performed by 4 different subjects. The camera is mounted on a cap worn by the subject.

Activity\Subject Subject 1 Subject 2 Subject 3 Subject 4
Cheese Sandwich S1_Cheese_C1 S2_Cheese_C1 S3_Cheese_C1 S4_Cheese_C1
Coffee S1_Coffee_C1 S2_Coffee_C1 S3_Coffee_C1 S4_Coffee_C1
Coffee with Honey S1_CofHoney_C1 S2_CofHoney_C1 S3_CofHoney_C1 S4_CofHoney_C1
Hotdog Sandwich S1_Hotdog_C1 S2_Hotdog_C1 S3_Hotdog_C1 S4_Hotdog_C1
Peanut-butter Sandwich S1_Peanut_C1 S2_Peanut_C1 S3_Peanut_C1 S4_Peanut_C1
Peanut-butter and Jam Sandwich S1_Pealate_C1 S2_Pealate_C1 S3_Pealate_C1 S4_Pealate_C1
Sweet Tea S1_Tea_C1 S2_Tea_C1 S3_Tea_C1 S4_Tea_C1

Download a video which can be used for calibrating the camera (C1_Calibration.MP4)

Download the code for extracting images from videos, calibration and resizing (extract images code), you need to have ffmpeg installed in order for this code to work.

In case you prefer to download the raw images instead of running the calibration and other code to extract them from videos, here is the link to raw images (raw images - 10GB).

Download the annotations (Labels)

Download the left hand, right hand and object segment masks computed by our algorithm (masks)

Cite the following paper when using the dataset: