Communication is quite a fundamental aspect for us humans to survive, grow, and excel. Today it is all the more relevant, with the whole world becoming one big habitat. There is nothing that shall stop us from understanding or conveying thoughts to anyone anywhere in this world. Owing to technology, travels, and communications are becoming more comfortable and more fun.

The global boundaries are shrinking as we all dream and aspire to travel more and more. Hence, we all need to be able to understand and overcome any language barriers. In India specifically, with 22 constitutionally recognised languages written in 13 different scripts, it becomes critical we come up with technological advancements that bridge any gap present. 

Google also transitioned to deep learning to improve its efficiency by around 60%. Today, Google and Microsoft can translate more than 100 languages and are aiming for human-like accuracy.

IIT-Madras scientists have launched AI4Bharat to boost AI innovation in India. AI4Bharat is a community of engineers, domain experts, policymakers, and academicians collaborating to build AI solutions to solve the problems very specific to India. The ongoing projects from AI4Bharat around NLP include Signboard Translation from Vernacular Languages, Fonts for Indian Scripts, Word embeddings for Indian Languages, and many more.

Talking about the signboard translation project, we can say that it is a wonderful effort in the direction. The idea is to create an application that uses NLP at its core to translate the text written on a signboard to another language as desired by the user.

The user just has to point towards a signboard using the camera from the device such as his/her phone or tablet, and the app will translate the text in the desired language. The project, in the initial stage, aims to work on names specifically like city names, road names, etc. They are usually small text pieces with just a few words but are critical to spot locations during travels. The later stages aim to increase the number of languages as well as the text complexity.

Text complexity means the longer texts like “Work in progress ahead, please slow down.” Every state highway, national highway, or within the city signboards cannot be readable or understandable for travellers from other states or countries. Even the whole advertising on hoardings in order to increase its reach needs to be made comprehensive for a non-native. Moreover, the purpose of travel isn’t fully met if you don’t understand the language and can’t read the signboards, and hence miss the essence.

Going to the details, we can definitely say this app would require clear boundaries of a signboard to decide as to which part needs transition or to see which part consists of the text. So we also need to standardize signboards that are compatible with global applications. The same applies to the text written inside the boxes, which must follow the syntax and semantics guidelines.

The outer boundary or the text box will help the system for detection and identification of the region containing the text. The text written inside will be used by engines as an input to convert in the desired language.

Some other usage areas that can be benefitted from this collaboration between language diversity and NLP are commerce, government, education, trade, investment, finance, and media. 

On these lines, technology companies are investing heavily in machine translation. Investments and advancements in deep learning are improving the accuracy and quality of these translations leading to more faith in NLP.

Tourism is a key sector in India, with India being culturally rich and the state governments are promoting the same. It requires a collaborative approach from tech companies and governments to achieve a more tourism-friendly environment. This signboard project shall be a milestone for a truly digital India.