nano-JEPA: Democratizing Video Understanding with Personal Computers

• Autores: Rostagno, Adrián; Iparraguirre, Javier; Ermantraut, Joel; Tobio, Lucas; Foissac, Segundo; Aggio, Santiago; Friedrich, Guillermo Rodolfo
• Año de publicación: 2024
• Idioma: inglés
• Tipo de recurso: documento de conferencia
• Materia: Ciencias Informáticas; feature prediction; unsupervised learning; visual representations; video; joint-embedding predictive architecture
• Institución: Universidad Nacional de La Plata

The Video Joint Embedding Predictive Architecture (V-JEPA) has shown great promise in self-supervised video representation learning. However, its substantial computational demands, often necessitates powerful GPU clusters, limit accessibility for many researchers. We introduce nano-JEPA, a streamlined adaptation of V-JEPA designed to run efficiently on resource-constrained personal computers, even those with only CPUs. Additionally, we present the nano-datasets repository, facilitating the creation of manageable subsets from large public video datasets. Our work aims to democratize research in this field, enabling broader participation and experimentation with V-JEPA-like models. We demonstrate that nano-JEPA, trained on smaller datasets and hardware, can still achieve reasonable performance on downstream tasks, opening doors for further exploration and innovation.
Red de Universidades con Carreras en Informática