Multimodal text summarization is a complex and challenging task in the field of natural language processing. Its objective is to use a combination of features from various modalities to create a concise yet informative summary from a given set of input data. In our research, we conducted a thorough survey of various techniques and methods used for multimodal text summarization and analyzed their implications on both research and industry. Moreover, we have developed a straightforward yet efficient model to address the challenges associated with this task. Our model has achieved state-of-the-art performance on the MMSS dataset. Additionally, we have proposed a semantic-based evaluation technique to measure the quality of the generated summaries. The effectiveness of our proposed technique has been substantiated through empirical evidence and appropriate analysis and discussion. Our goal is to make the code and models we have developed for our system publicly accessible.