SIGNO-PERU-FER: A Multimodal RGB-D Dataset of Visuo-Gestural Facial Expressions in Peruvian Sign Language

SIGNO-PERU-FER is a multimodal RGB-D dataset of dynamic Visuo-Gestural Facial Expressions (VGFEs) in Peruvian Sign Language (LSP), designed to support research on non-manual sign language recognition, multimodal facial behavior modeling, and inclusive AI systems. It comprises 1,256 synchronized RGB-D video sequences from 16 native LSP signers across 14 VGFE categories selected for their visual similarity and classification challenges. Each sequence includes RGB video (1920×1080, 30 fps), depth maps (512×512), and 468 3D facial landmark points captured with an Azure Kinect DK under controlled laboratory conditions. # Category # Category 1 Annoyed 8 No 2 Bored 9 Sad 3 Crying 10 Smells bad 4 Disgust 11 Surprise 5 Happy 12 Thief 6 I don't know 13 Tired 7 Let's see 14 Yes The SIGNO-PERU-FER dataset is archived in Zenodo under restricted access due to the re-identification risk inherent to audiovisual recordings. Access is provided via Zenodo's access-request workflow and requires acceptance of the SIGNO-PERU-FER Data Use Agreement (DUA), which includes named-user registration and data protection terms. To request access, download the SIGNO-PERU-FER_AccessRequestForm_DUA from this record, complete it, and email the signed form to [e-mail] or [e-mail].