For the laboratory of Psychophysics, EPFL, Switzerland
This project aims at modelling visual grouping with the Sequential Metacontrast Paradigm (a vision paradigm).
- Different data sets tested.
- Different models of reconstructors (LSTM, PredNet, ...) and decoders tested (Dense, fully-convolutionnal, ...).
- Different criteria, like minimum entropy, tested to model the "conscious percept" formation by the visual system and its discrete behavior.
The final project report is in sqm_models_Marie_Anselmet.pdf