Combating the hate speech in Thai textual memes

Abstract

<span>Thai textual memes have been popular in social media, as a form of image information summarization. Unfortunately, many memes contain some hateful content that easily causes the controversy in Thailand. </span><span>For global protection, t</span><span>he </span><em><span>Hateful Memes Challenge</span></em><span> is also provided by </span><em><span>Facebook AI</span></em><span> to enable researchers to compete their algorithms for combating the hate speech on memes as one of </span><em><span>NeurIPS’20</span></em><span> competitions. As well as in Thailand, this paper introduces the Thai textual meme detection as a new research problem in Thai natural language processing (Thai-NLP) that is the settlement of transmission linkage between scene text localization, Thai optical recognition (Thai-OCR) and language understanding. From the results, both regular and irregular text position can be localized by one-stage detection pipeline. More scene text can be augmented by different resolution and rotation. The accuracy of Thai-OCR using convolutional neural network (CNN) can be improved by recurrent neural network (RNN). Since misspelling Thai words are frequently used in social, this paper categorizes them as synonyms to train on multi-task pre-trained language model. </span>

Automated Data Digitization System for Vehicle Registration Certificates Using Google Cloud Vision API

“…It is consistent with other studies such as Vaithiyanathan & Muniraj [12], Zaki et al [23] and Thammarak et al [18]. In addition, Thai characters that often result in error include headless characters, such as ก and ธ, and letters with slightly different shapes, such as ด and ตand ฎ and ฏ, which is consistent with the study of Somboonsak [17], Mookdarsanit & Mookdarsanit [15] and Thammarak et al [18].…”

“…For example, Chumwatana and Rattana-Umnuaychai [14] proposed OCR technology to recognise Thai text from a physical document in a digital format. Mookdarsanit & Mookdarsanit [15] presented a combination of hate speech in Thai text memes using Thai OCR. Furthermore, Chomphuwiset [16] presented a technique for recognising printed Thai characters using a feature-based technique and a convolutional neural network (CNN).…”

“…Many studies in Thailand have focused on applying optical character recognition (OCR) [14,15] or machine learning technology [16]; however, findings demonstrate that Thai character recognition is still less effective than that of English characters. Somboonsak [17] found detecting new words that were not in the dictionary and adding new words to the dictionary difficult.…”

Thai Text-to-Image Prompt Engineering by Pre-Trained Large Language With Stable Diffusion Model

“…In the deep learning age, Thai-OCR had no more challenges (recognizing Thai characters from the image and understanding semantic text). Thai-OCR was expanded to meme image categorization (Mookdarsanit & Mookdarsanit, 2021a). Or scene text detection (Kobchaisawat, Chalidabhongse & Satoh, 2020).…”

“…As motivated by Thaidessert image synthesis (Mookdarsanit & Mookdarsanit, 2018d)., TH-T2I could be further enhanced in many semantic Image-Text relations in Thai culture and art (Mookdarsanit & Rattanasiriwongwut, 2017c). Available on large-scale social media: street surveillance (Sutthaluang & Prakancharoen, 2020)., plant recognition (Mookdarsanit & Mookdarsanit, 2019b)., image location estimation (Mookdarsanit & Rattanasiriwongwut, 2017b)., Buddhism and temples (Mookdarsanit & Rattanasiriwongwut, 2017a)., food image description (Soimart & Mookdarsanit, 2017a)., tourism classifi cation (Mookdarsanit & Mookdarsanit, 2018c)., GPS place estimation (Soimart & Mookdarsanit, 2017b), pesticide analytics (Sutthaluang, 2019)., agricultural product quality (Mookdarsanit & Mookdarsanit, 2021a)., facial verifi cation and recognition (Soimart & Mookdarsanit, 2016)., and cosmetic recommender systems (Mookdarsanit & Mookdarsanit, 2023).…”