Text or Non-text Image Classification using Fully Convolution Network (FCN)
Neeraj Gupta, Anand Singh Jalal · 2020
The semantic information in a natural scene plays a vital role in image understanding. One of the semantic information present in a natural image is the text. It can be utilized for analyzing several computer vision applications. The proposed work focuses on the new task of classifying the text images from a bulk of natural images. A model is intended to address this issue; it uses a lightweight, fully convolution network (FCN) which classifies the image as the text or non-text image. The effectiveness of the proposed model is demonstrated on the standard dataset, experimental evaluation and comparison with the baseline method show that the proposed model outperforms.