A Long Skip Connection for Enhanced Color Selectivity in CNN Architectures

Some recent studies show that filters in convolutional neural networks (CNNs) have low color selectivity in datasets of natural scenes such as Imagenet. CNNs, bio-inspired by the visual cortex, are characterized by their hierarchical learning structure which appears to gradually transform the repres...

Descripción completa

Detalles Bibliográficos
Autores principales: Sanchez-Cesteros, Oscar, Rincon, Mariano, Bachiller, Margarita, Valladares-Rodriguez, Sonia
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2023
Materias:
Article
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10490730/
https://www.ncbi.nlm.nih.gov/pubmed/37688036
http://dx.doi.org/10.3390/s23177582

Internet

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10490730/
https://www.ncbi.nlm.nih.gov/pubmed/37688036
http://dx.doi.org/10.3390/s23177582

Ejemplares similares