Skip to the content.

图片

Introduction

HECO consists of images from the HOI datasets, film clips, and images from the Internet. The database contains a total number of 9,385 images and 19,781 annotated agents. Such image combination contains rich context information and diverse agent interaction behaviours.

Annotation

For discrete categories, we annotate with eight categories, including Surprise, Excitement, Happiness, Peace, Disgust, Anger, Fear, and Sadness.


Examples of the recognized agents with different discrete emotion categories are included in the HECO.


For continuous dimensions, we use the emotional state model of VAD, and annotate the Valence (V), Arousal (A) and Dominance (D) of agents on a scale of 1-10.


Examples of the recognized agents with different scores of Valence, (row 1), Arousal (row 2) and Dominance (row 3) are included in the HECO.


Besides, we enforce numerical values to express relative percentages. Each category’s count and the distribution of continuous dimensions across different categories are shown as follows.


Count and per each continuous dimension’s distribution of the scores across the different categories.


Inspired by emotion sociology, we propose two novel label spaces : Self-Assurance (Sa) and Catharsis (Ca). Sa refers to the level of confidence in the agent’s ability and judgement, i.e., the agent conveys feelings of competence and adequacy, representing the degree to which the agent understands emotion at the cognitive level. Ca reflects the influence of change in emotion from the agents on interaction and situation.

(a) Distribution map of Sa. (b) Distribution map of Ca.

Download the HECO Dataset

Researcher should use HECO only for non-commercial research and educational purposes.

The dataset (images and annotations) is avaliable at  here.

Acknowledgements

This work is supported by National Key R&D Program of China (2021ZD0113502, 2021ZD0113503), Shanghai Municipal Science and Technology Major Project (2021SHZDZX0103) and National Natural Science Foundation of China under Grant (82090052)