Multimodal Retrieval for Semantic Matching and Semantic Correlation Matching through Different Classification Models

Now a day multimodal retrieval is an important research topic in computer vision. Due to its powerful application in research fields such as image retrieval, text retrieval, video retrieval, identifying defective people, human-machine interaction or robotics, and so on. Image to text and text to image is a cross-modal retrieval system. In this paper, we try to introduce Scale-invariant Feature Transformation (SIFT) feature-based different classification techniques such as Kernel Extreme Learning Matching (KELM), Support Vector Matching (SVM) with semantic matching (SM), and semantic correlation matching (SCM) and correlation matching (CM). Kernel Extreme Learning Matching is newly classifier in this field. By using this classifier, we get consistent performance in the case of image to text and text to image queries. We compare the mean average precision (mAP) in both SM and SCM cases with Support Vector Machine (SVM) and Kernel Extreme Learning Matching (KELM) classification techniques. Finally, we observe that KELM-based SM with Centered Correlation performs well and more accurate results.

• Supervision