Data Analytics for Pandemics by Shinde Gitanjali Rahul; Kalamkar Asmita Balasaheb; Mahalle Parikshit N

Author:Shinde, Gitanjali Rahul; Kalamkar, Asmita Balasaheb; Mahalle, Parikshit N.
Language: eng
Format: epub
Publisher: Taylor & Francis Group

2.4.1.2 Knowledge Extraction from Image Data

Due to the sudden rise in social media use and globalization, the production of image data has increased. This type of data is visually very rich, but it is in a structured format. Extracting meaningful semantic and visual construct is one of the challenging tasks. In KE, image feature extraction, understanding of the scene, and text recognition are some very important areas to focus on. In order to find out objects in the given image, feature extraction can be helpful. Along with feature extraction, different classification and segmentation methods can be used for the same purpose. Feature extraction from an image can be done with the help of the Scale Invariant Feature Transform (SIFT). In this method text resides inside the image and this type of knowledge is generally very large. Text extraction methods are used in order to extract meaningful knowledge from the image. But this task comes with its own set of challenges--it is very difficult to find out the text due to different languages, sizes, orientation, contrast, background colors, etc. Various approaches have been designed for this purpose, but these approaches have limitations like domain or language. For the extraction of text from an image, no single technique is available that can be implemented in all scenarios. The benefits of KE from images are high accuracy and less complexity. However, if the image in question contains noise then that noise must be removed before the implementation of KE. To get a better understanding of the visual/scene, a top-down, bottom-up, or combined approach is used. In real-time or dynamic scenarios, a full understanding of the scene and the static and dynamic occlusions makes KE very challenging.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.