Communication Dans Un Congrès Année : 2025

AMDA: Advancing Multimedia Data Annotation for Human-Centric Situations

Ibrahim Serouis

Résumé

Recent strides in AI research, particularly in computer vision and natural language processing, have significantly advanced the partial automation of data labeling and annotation processes. However, there remains a notable void in applying these cutting-edge techniques to videos portraying human-centric scenarios, with scant exploration of automated solutions for multimedia data. Current research primarily focuses on visual cues, such as on-screen detections, textual cues such as named entity recognition, and auditory cues involved in speech-to-text conversion. This paper proposes a methodology that leverages state-of-the-art deep learning techniques to extract multimedia cues from videos. Through evaluation across various video contexts, our methodology yields promising results, potentially charting a course for future research endeavors.

Fichier non déposé

Dates et versions

hal-04909771 , version 1 (24-01-2025)

Identifiants

Citer

Ibrahim Serouis, Florence Sedes. AMDA: Advancing Multimedia Data Annotation for Human-Centric Situations. MultiMedia Modeling 2025, Jan 2025, Nara Japan, Japan. pp.84-90, ⟨10.1007/978-981-96-2074-6_7⟩. ⟨hal-04909771⟩

Collections

80 Consultations
0 Téléchargements

Altmetric

Partager

  • More