Journal: Frontiers in Neuroinformatics

Article Title: An action decoding framework combined with deep neural network for predicting the semantics of human actions in videos from evoked brain activities

doi: 10.3389/fninf.2025.1526259

Figure Lengend Snippet: Overview of the proposed method. (a) The X3D model. The orange arrows represent the learning process of the deep learning model X3D from video input to output action semantics. (b) The MLP regression model. The blue arrows represent the process of predicting action features from fMRI, and then from the predicted features to action semantics. Images were taken from (Creative Commons License CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: Imaging data are collected using a 3T Siemens Prima functional magnetic resonance imaging scanner.

Techniques:

Journal: Frontiers in Neuroinformatics

Article Title: An action decoding framework combined with deep neural network for predicting the semantics of human actions in videos from evoked brain activities

doi: 10.3389/fninf.2025.1526259

Figure Lengend Snippet: Top-5 estimated semantic categories of human action videos. The vertical line shows the top-5 categories determined from fMRI activity which are shown in the order of descending probabilities from the top to the bottom. The horizontal one shows predicted probability of estimated categories, and correct categories are highlighted in red. Images were taken from (Creative Commons License CC BY 4.0, https://creativecommons.org/licenses/by/4.0/ ).

Article Snippet: Imaging data are collected using a 3T Siemens Prima functional magnetic resonance imaging scanner.

Techniques: Activity Assay