Image Data Collection for Text Recognition Model

Our company has been approached by one of our esteemed Fortune 500 clients who has expressed an interest in developing a Thai language-specific model. The aim of the model is to detect and recognize multilanguage text within images in real-time and offline use cases. Specifically, they have requested text recognition capabilities to be added to their camera, enabling users to detect, translate, share, and copy text directly from the camera screen.

To facilitate this project, we have worked collaboratively with our client and assisted them in collecting more than 40,000 images across various categories. As we are all aware, the field of machine learning (ML) is continuously evolving and experiencing numerous breakthroughs, leading to the development of models for new use cases. However, it is important to note that the success of an ML model is dependent on the amount and quality of training data used. Therefore, each use case has specific data requirements to make ML models more reliable and interactive with users in real life.