Extractive Summarization of Text from Images
Raunak Kolle, S Sanjana, Merin Meleet · 2021 International Conference on Innovative Computing, Intelligent Communication and Smart Electrical Systems (ICSES) · 2021
The modern age has brought with it an abundance in the inflow of information. The consumption of such large chunks of information leads to a latency in the gathering of relevant information. The condensation of such information becomes necessary as the volume of this inflow keeps expanding. The most efficient way for the retrieval of the most important contents of the data is to summarize the data such that only non-redundant and useful information is contained in it. The manual summarization of textual content is a laborious task and is not very efficient. Therefore, automatic summarization is of utmost importance. Text summarization is the process of identifying the most relevant information and discarding the unnecessary and the irrelevant information. In this paper, textual content is extracted from images using optical character recognition and the extracted text is subjected to various extractive summarization algorithms. The summary is evaluated with respect to the summary generated by an abstractive summarization model using Rouge-N and Rogue-L metric to calculate the Precision, Recall and F-Score.