Reproduce “AVID: A 5T fMRI Dataset for Benchmarking Auditory-induced Visual Mental Imagery Decoding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shiqi Shen, Shurui Li, Xilin Zhang, Yuanning Li
AVID presents a benchmark for decoding visual mental imagery from spoken Mandarin descriptions, without showing participants source images during imagery. It records about 67 hours of fMRI from nine participants, across natural-scene and controlled-attribute splits. Existing decoders are evaluated for source-image consistency and description alignment. MindEye1 reaches 67.83% CLIP-T accuracy on the complex split, above 50% chance but not evidence of exact mental-image recovery.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.