Ego4D V2 Suggestions

We’re actively planning for the next version of Ego4D - significantly more hours collected across modalities and with multiple camera views (calibrated and sync’d), the next iterations of annotations (extensions of the current ones and entirely new categories), and a far more complete taxonomy/hierarchy/structure. We’re also looking to significantly expand the features, models and utilities available with the dataset.

We’d love to hear your wish list! (Or v1 complaints that could be corrected.)

Please start a new topic for each new category or distinct suggestion, and don’t be shy, even if it’s narrow it very well might fit into roadmap…

Thank you for the dataset!

The lack of camera intrinsics makes any 3D analysis on the data very hard. Even approximate camera intrinsic will make working with the data much easier.

I hope to see it in the next iteration of the dataset :slight_smile: