Deep Learning based Automatic Image Annotation System for Image Retrieval with Arabic language

Résumé

Abstract Social media platforms like YouTube, Twitter, and Facebook have grown into new modalities of communication, allowing a great number of individuals to interact and learn. Furthermore, many social media users currently produce and share incorrect thoughts and images that are not accompanied by words. The automated generation of any text for raw photos is a difficult operation, particularly when working with Arabic and a limited amount of training samples. To tackle this challenge, smart technology and deep learning technology have been offered. The method employs a bi-level architecture, which allows for the use of self-supervision to produce rotation variants in order to increase the number of training samples. This, in turn, enhances the variability of the model representation and enables the investigation of high-level object information for more detailed picture production. GoogleNet model automatically overcomes the stability issues associated with parsing and synthesising any picture. The proposed technique makes use of deep learning technology to explain images received from the internet environment and gather information about the existence of identified items in order to create an optimal detection method that does not interfere with the deep learning process.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0