|
09 May 2023, Volume 38 Issue 3
|
|
|
Abstract
Thoughts, feelings, desires, or views can be expressed using words, photographs, or moving pictures. A developing topic of social analytics study is sentiment analysis of web content. Using various social networking sites like Instagram, Linkedin, Twitter, Messenger, and others, users communicate their feelings online by sharing texts and sharing photographs. There have been little studies on sentiment analysis of visual information; nevertheless, there's been a huge amount of research on sentiment analysis of textual information. The effectiveness of Convolution Neural Networks (CNNs) for object detection and verification utilizing various networks is examined in this research. There are numerous networks available to test how well the various networks operate. Convolution Neural Network results are evaluated using the well-known standard dataset ImageNet as well some real time image dataset. This research focuses on the investigation of three well-known networks, VGG16, VGG19, ResNet50 on ImageNet dataset. These three models are implemented using Keras & Tensorflow for picture categorization. These models are also used to classify and assess the correctness of an arbitrary collection of tagged photos from the web. In comparison to VGG16 and VGG19, it has been found that ResNet50 can detect images more accurately.
Keyword
Image Sentiment Analysis, VGG16, VGG19, ResNet50, Convolutional Neural Network
PDF Download (click here)
|