Priyanjali Gupta, a talented computer science student, built a model to translate sign language in real-time

A little data labeling and literally 3 lines of code.

Priyanjali Gupta is an engineering student who developed an AI model that translates American Sign Language (ASL) into English in real-time. She was a third-year computer science student specializing in data science at the Vellore Institute of Technology, Tamil Nadu, when she created this model. Her work was inspired by a challenge from her mother to use her engineering studies to create something impactful. The AI model uses Tensorflow's object detection API and transfer learning from a pre-trained model to detect six ASL signs: Hello, I Love You, Thank you, Please, Yes, and No

Copy Pasta for every engineering student. xD

This is a great initiative, but students in the Computer Science and Engineering field will recognize that this is a very basic machine learning model, typically performed by third and fourth-year students as part of their curriculum.

That is a very cool project for a student. There are many approaches to do this, ranging from fine tuning a yolo algorithm to using a two stages inference with preexisting model like mediapipe's models offered by Google following by a classifier. These things are now basic. You can do this now with a multimodal LLM if you want :) and the same model will also answer suestions and even convert the images to text. Ai of tuday is very advanced although we just scratched the surface.

Eventually, these will be fit to contact lens or glasses. And the reverse will happen too - people talking will have visual VR hands to translate their normal speech to a deaf person. Amazing technology.

Can this system detect fast hand movements and unite several gentures and make sentences? If yes, then this is a great system, if not this is still bad realization and needs a lot of improvements before being released.

Hundreds of open source projects are on GitHub. It needs almost no efforts to make a demo like this. 🥲