The Multi-PIE (Multi Pose, Illumination, Expressions) dataset consists of face images of 337 subjects taken under different pose, illumination and expressions. The pose range contains 15 discrete views, capturing a face profile-to-profile. Illumination changes were modeled using 19 flashlights located in different places of the room.
Source: Hybrid VAE: Improving Deep Generative Models using Partial Observations
Variants: Multi-PIE
This dataset is used in 1 benchmark:
No recent benchmark submissions available for this dataset.
No papers with results on this dataset found.