The impact of expertise on brain computer interface based salient image retrieval

Area of Science:

• Computer Vision
• Neuroscience
• Human-Computer Interaction

Background:

• Autonomous decision-making in computer vision relies on accurate object and event recognition.
• Training data is crucial for developing robust computer vision algorithms, but creating salient image datasets is challenging.
• Electroencephalogram (EEG) and brain-computer interface (BCI) offer potential for objective measurement of visual saliency.

Purpose of the Study:

• To develop a new approach for creating training datasets for computer vision algorithms.
• To investigate the use of EEG and BCI signals for retrieving salient images.
• To explore the influence of subject expertise on the image retrieval process.

Main Methods:

• Utilized rapid image presentation combined with EEG and BCI to capture brain responses.
• Recruited two groups of participants: experts and novices in the relevant domain.
• Analyzed EEG signals to differentiate brainwave patterns associated with salient image perception.

Main Results:

• Achieved relatively high accuracy in retrieving salient images across most participants.
• Demonstrated a clear distinction in brainwave patterns between expert and novice subjects when viewing salient images.
• Confirmed the feasibility of using EEG/BCI for objective saliency-based image selection.

Conclusions:

• EEG and BCI provide a viable method for generating high-quality training data for computer vision.
• Subject expertise significantly impacts neural responses to visual stimuli, affecting saliency detection.
• This approach enhances the development of more intelligent and context-aware computer vision systems.