Research on Image Sentiment Analysis Based on Feature Pyramid and Attention Mechanism

Sentiment analysis of images allows for a deeper understanding of the information conveyed by the images, better meeting people's needs in visual content comprehension and interaction. This paper proposes a model that integrates ResNet50 with a feature pyramid and attention mechanism. The model first improves upon the traditional ResNet50 by incorporating a feature pyramid module, which integrates multi-scale feature maps and enhances the accuracy of image sentiment analysis. Then, the CBAM attention module is introduced to enhance attention to key features. The model integrates both channel and spatial attention mechanisms to effectively capture key channel and spatial details in images. The proposed image sentiment analysis model, which incorporates feature pyramids and attention mechanisms, demonstrates high accuracy and reliability in feature extraction and sentiment analysis tasks.