An EEG-based Image Annotation System

TL;DR

This paper introduces an EEG-based image annotation system that leverages brain signals to rapidly and accurately label images, significantly increasing annotation throughput without requiring category-specific training.

Contribution

It presents a novel EEG-driven annotation method using P300 signals during RSVP tasks, achieving high accuracy and fast labeling without prior category-specific models.

Findings

Achieves up to 10 images per second annotation rate

F1-score of 0.88 on test set

Does not require pre-trained models for new categories

Abstract

The success of deep learning in computer vision has greatly increased the need for annotated image datasets. We propose an EEG (Electroencephalogram)-based image annotation system. While humans can recognize objects in 20-200 milliseconds, the need to manually label images results in a low annotation throughput. Our system employs brain signals captured via a consumer EEG device to achieve an annotation rate of up to 10 images per second. We exploit the P300 event-related potential (ERP) signature to identify target images during a rapid serial visual presentation (RSVP) task. We further perform unsupervised outlier removal to achieve an F1-score of 0.88 on the test set. The proposed system does not depend on category-specific EEG signatures enabling the annotation of any new image category without any model pre-training.