The Diverse Single-eye Event-based Emotion dataset, DSEE

Yang Wang, Haiyang Mei, Qirui Bao, Ziqi Wei, Mike Zheng Shou, Haizhou Li, Bo Dong, Xin Yang.

Apprenticeship-Inspired Elegance: Synergistic Knowledge Distillation Empowers Spiking Neural Networks for Efficient Single-Eye Emotion Recognition. IJCAI 2024

IJCAI2024_MSKD.pdf

Please cite this paper when using the data： Bib.txt

Introduction:

DSEE contains intensity video frames and corresponding real/synthetic events as well as a ground truth emotion label. To the best of our knowledge  ,  DSEE is currently the largest single-eye event-based emotion benchmark  (kindly see Table 1 for a summary and Figure 1 for representative examples).

Table 1: Comparison among event-based datasets for emotion recognition.

Figure 1: Examples from our DSEE dataset.

As illustrated in the steps shown in Figure 2, we can obtain the intensity sequence, the corresponding events sequence, and the emotion label triplets, covering a diverse range of examples.

Figure 2: Illustration of the single-eye events data synthesis process.