Intelligent Image Text Reader using Easy OCR, NRCLex & NLTK

The Intelligent Image Text Reader is capable of capturing an image and extracting the text from the image using the Easy OCR library and displaying them in the form of speech output. The Text to Speech conversion methodology was earlier useful for the visually impaired but now it has got a wider scope. In the era of Digitalization, text is seen almost on all the online web sites, blogs, and e-learning materials. Text plays a vital role in some of the real-time applications like the vehicle license plates, traffic sign boards, banners and so on. Our proposed methodology is not just capable of identifying the text in the images but is also capable of translating the text from a common language to one's own native language with the help of Google Translate, additionally these texts are converted to speech using TTS (Text-to-Speech). This feature is helpful for a set of people who have low literacy levels, those who have an interest in learning a new language, and also for elderly people having low eyesight. The proposed method also helps in identifying a person's emotion based on the text and emojis sent during a conversation. The emotional intelligence of a person is identified using NRCLex (lexicon-based method) in our proposed solution. This feature is mainly helpful in fields like marketing, wherein, the customer support agent has a conversation with the customer in order to identify their customer satisfaction. The growing generation is much addicted to the social media networks like WhatsApp, Facebook, tweeter etc., With the help of emotion detection characteristic we can easily identify the public opinion regarding a particular issue, rate of stress & anxiety prevailing among the young generations. Also, the NLTK based method helps in removing the Stop words and help in identifying the Synonyms & Antonyms of a word which is a very useful feature for reading any document with correct vocabulary and understanding. The accuracy of this model is 98% approximately.