Özet:

Abstract Nowadays, online reviews and rating has become a powerful means of sharing opinions among the customers. From past few years people have started reacting about restaurants through pictures as reviews. The facial expression of the person or food in the image is reviewed along with a short text added to it and then classified into positive or negative category. This gives a clear understanding of the image. Because analysis of text sentiment is carried over enormously in natural language processing(NLP), this work focuses on visual sentiment analysis of restaurant review image dataset to find if the image belongs to positive or a negative category. An image prediction model is built using deep learning method such as Convolutional Neural Networks(CNN) to identify the sentiment. Image classification is performed and accuracy is enhanced using images posted on social media sites. The proposed system performs comparatively better than the machine learning methods in analyzing the opinions in images.

Anahtar Kelimeler: