Brain2Face
Brain2Face: A Study on the Relation of EEG Signals and Spatio-temporal Head Avatars
A dataset and benchmark protocol evaluating the recovery of 3D Gaussian head avatar expression latents from 64-channel EEG signals.
How to run
$ npm i -g lettokenburn-cli$ ltb bench info brain2face
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
9 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| REVE | 0.7 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| CodeBrain | 0.7 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| EEGPT | 0.9 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| BENDR | 0.9 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| CBraMod | 0.9 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| Luna | 1.0 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| Signal-JEPA | 1.0 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| BIOT | 1.1 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
| LaBraM | 1.3 | N/A | Independent | medium | Oct 12, 2026 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsAbout
The benchmark features paired 64-channel EEG recordings and spatio-temporal head geometry/appearance annotations encoded as latent expression parameters of personalized Gaussian avatars. It introduces an evaluation protocol and baseline models to measure the accuracy of regressing facial expressions from neural signals.