• KOTHAPALLI DEVIKA
• DR.CH.AVINASH

In our increasingly interconnected digital world, gaining a profound understanding of human emotions and concepts is indispensable for making insightful decisions across diverse fields. In this sentiment analysis plays a vital role, enabling us to go through the digital interactions by decoding the emotional content expressed. However, traditional approaches often face challenges when confronted with the intricacies of human language, as they overlook crucial information conveyed through visual means. To overcome this constraint, we introduce an innovative method that utilizes the combination of multi-modal data fusion and deep learning approaches. We employ a hybrid model design to integrate textual and visual data. We achieve this by using pre-trained TF-IDF to extract textual features and finetuned ResNet50 to extract visual features. We fuse the features and then perform sentiment classification. Through thorough experimentation, we demonstrate the effectiveness of our method, especially in situations where textual and visual signals provide additional information. The outcomes of our study exhibit substantial progress in both the accuracy and reliability of sentiment classification when compared to conventional unimodal methods. Furthermore, we use our algorithm on actual data, effectively forecasting emotions conveyed in both textual and visual formats. The assessment, which utilizes classification metrics, showcases the exceptional accuracy of our multimodal fusion technique, achieving 98% accuracy. This research enhances sentiment analysis approaches by leveraging the combined strength of textual and visual data. The approach we propose presents a great opportunity to improve sentiment comprehension in diverse areas such as social media analysis, customer feedback interpretation, and opinion mining. We are confident that our study will provide stakeholders with practical and valuable information obtained from the combination of written and visual signals. This will facilitate more knowledgeable decision-making in the digital era.

Image Captioning, Feature Extraction, Deep Learning, ResNet50, BLIP Model, Sentiment analysis, Text features, Image features, Modalities, Multi-modal Data Fusion.

"EXPLORING DEEPLEARNING APPROACHES FOR MULTIMODAL DATAFUSION IN SENTIMENT ANALYSIS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 6, page no.g88-g94, June-2024, Available :http://www.jetir.org/papers/JETIR2406615.pdf

"EXPLORING DEEPLEARNING APPROACHES FOR MULTIMODAL DATAFUSION IN SENTIMENT ANALYSIS", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 6, page no. ppg88-g94, June-2024, Available at : http://www.jetir.org/papers/JETIR2406615.pdf