|Description (include details on usage, files and paper references)||EPIC-KITCHENS, is the largest egocentric video benchmark recorded by 32 participants in their native kitchen environments. Our videos depict non-scripted daily activities, captured using a head-mounted camera (Full HD, 60fps). Recording took place in 4 cities (in North America and Europe) by participants belonging to 10 nationalities.
EPIC-KITCHENS consists of 11.5M frames, which we densely labelled for a total of 39.6K frame-level action segments and 454.2K object bounding boxes. Our annotation is unique in that we had the participants narrate their own videos (after recording), thus reflecting true intention, and we crowd-sourced ground-truths based on these.
We describe our object detection, action recognition and action anticipation challenges, and report baselines in two scenarios; seen and unseen kitchens. We released all data and training set annotations, and will soon track the communitys progress on all challenges (with held out test ground-truth) via an online leaderboard.
Full details available in the Arxiv paper: https://arxiv.org/abs/1804.02748
A video introduction to EPIC-KITCHENS is available at: http://youtu.be/Dj6Y3H0ubDw
Dataset and annotations available from: http://epic-kitchens.github.io/
For inquiries: firstname.lastname@example.org
Dima Damen (1) Hazel Doughty (1) Sanja Fidler (2) Giovanni Maria Farinella (3) Antonino Furnari (3) Evangelos Kazakos (1) Davide Moltisanti (1) Jonathan Munro (1) Toby Perrett (1) Will Price (1) Michael Wray (1)
(1 University of Bristol)
(2 University of Toronto)
(3 University of Catania)