Tldsmi: Genetic Algorithm Based Network for Text Localization in Distorted Social Media Images

Palaiahnakote, Shivakumara; Pavan Kumar, C.; Aggarwal, Pranjal; Sharma, Shubham; Chandana, Pasupuleti; Basavanna, M.; Pal, Umapada

Abstract

This paper presents a novel model for understanding social image content through text localization. For text localization, we explore Maximally Stable Extremal Regions (MSER) for detecting components, that works by clustering pixels having similar properties. The output of component detection includes several non-text components due to degradations of social media images. To select the best components among many, we explore Genetic Algorithm by convolving different kernels with components, which results in a feature matrix which is further fed to EfficientNet for choosing actual text components. Therefore, the proposed model is called Genetic Algorithm based Network for Text Localization in Distorted Social Media Images (TLDSMI). For evaluating text localization, we consider the images of standard dataset of natural scene by uploading and downloading from different social media platforms, namely, WhatsApp, Telegram and Instagram. The effectiveness of our method is shown by testing on original and distorted standard datasets.

Citation

Palaiahnakote, S., Pavan Kumar, C., Aggarwal, P., Sharma, S., Chandana, P., Basavanna, M., & Pal, U. Tldsmi: Genetic Algorithm Based Network for Text Localization in Distorted Social Media Images

• Working Paper Type: Working Paper
• Deposit Date: Nov 15, 2024
• Related Public URLs: https://dx.doi.org/10.2139/ssrn.4348525