Abstract
MoCA is a bi-modal dataset in which we collect Motion Capture data and video sequences acquired from multiple views, including an ego-like viewpoint, of upper body actions in a cooking scenario. It has been collected with the specific purpose of investigating view-invariant action properties in both biological and artificial systems. Besides that, it represents an ideal test bed for research in a number of fields – including cognitive science and artificial vision – and application domains – as motor control and robotics. Compared to other benchmarks available, MoCA provides a unique compromise for research communities leveraging very different approaches to data gathering: from one extreme of action recognition in the wild – the standard practice nowadays in the fields of Computer Vision and Machine Learning – to motion analysis in very controlled scenarios – as for motor control in biomedical applications. In this work we introduce the dataset and its peculiarities, and discuss a baseline analysis as well as examples of applications for which the dataset is well suited.
Highlights
Background & SummaryThe Multiview Cooking Actions dataset (MoCa) is a bi-modal dataset acquired to understand motion recognition skills and view-invariance properties of both biological and artificial perceptual systems.Unlike other recently proposed datasets, where actions and activities are observed in highly unconstrained scenarios[1,2], our dataset has been acquired in a set-up designed to achieve a compromise between precision and naturalness of the movement
Our dataset provides a collection of daily life activities, which could serve both in the context of action recognition from the robot camera and in the perspective of generating appropriate robot motions
We identified meaningful cut points of the marker position along the most significant axis with respect to the MoCap reference system, marking the end of the portion associated with an action instance
Summary
The Multiview Cooking Actions dataset (MoCa) is a bi-modal dataset acquired to understand motion recognition skills and view-invariance properties of both biological and artificial perceptual systems. Unlike other recently proposed datasets, where actions and activities are observed in highly unconstrained scenarios[1,2], our dataset has been acquired in a set-up designed to achieve a compromise between precision and naturalness of the movement Such properties make our dataset an ideal test bed for a number of fields and related research questions, among which it is worth mentioning the following:. Collaborative robotics, where a fast comprehension of what the partner is doing and when it is the right moment to act is a fundamental ability In this respect, our dataset provides a collection of daily life activities, which could serve both in the context of action recognition from the robot camera and in the perspective of generating appropriate robot motions
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
