You can create your account and contribute without agreeing to this.
Why this helps. The keypoints are what turn your video into training data a robot can actually learn from. They describe the movement in the footage, not who you are, so agreeing adds real value to the project at no risk to you.
What this is. A pose-estimation step computes the two- and three-dimensional coordinates of the joints of your body and of each of your hands — your skeletal keypoints — frame by frame. They are derived from your own body as it appears in the footage you film.
What it is used for. The keypoints describe the action being performed in the video, which is what makes the footage usable for robotics training. They are not used, and are not technically configured, to identify you.
Who receives it. Extraction runs on EgoVista’s EU-region infrastructure. The resulting keypoint annotations are included in the annotated dataset delivered to the buyer of that dataset. Buyers never receive your name, your email, your IP address, or anything else linking the footage to you.
How long we keep it. For the lifetime of the dataset version the annotations form part of. You can withdraw your consent at any time, in one click, from your account settings; we then stop extracting keypoints from your footage.
Saying no costs you nothing. If you decline, you can still join every campaign, you are paid exactly the same, and your account is unaffected. The only difference is that the keypoint extraction step is skipped on your footage. This choice is entirely yours, and you can change it at any time.