Success Story of Intel® Student Ambassadors Leveraging AI Tools from Intel
In a post-pandemic world where remote work and virtual experiences are increasingly common, the ability to communicate with individuals with hearing disabilities has become essential. This underscores the urgent need for a platform that teaches sign language in various dialects, facilitating communication across cultures and borders. Surprisingly, many video conferencing, e-learning, and entertainment platforms lack this feature. The ‘ASL Bridgify’ project aims to bridge this gap and revolutionize American Sign Language (ASL) education. Developed by a few Intel® Student Ambassadors and their team at the University of California, Berkeley’s AI Hackathon (held on June 22-23,2024), ASL Bridgify enhances communication and education among individuals with hearing disabilities. The AI solution leverages AI tools and frameworks optimizations on Intel® Tiber™ Developer Cloud.
About The Intel® Student Ambassadors
Prajwal Sharma is currently pursuing B.S. in Computer Science at Cal State University, Long Beach (CA). He is the founder and CEO of a startup called ‘Bazalu’. He is also the founder of a peer-to-peer private tutoring platform called ‘Better Grades Effortlessly’
Aahil Ali is currently pursuing B.S. in Computer Science at University of California, Los Angeles (UCLA).
Sapana Dhakal is a student at Cerritos College (CA).
Each one of the above leverage Intel Tiber Developer Cloud and oneAPI tools and frameworks to develop optimized AI solutions in collaboration with developer communities as part of the Intel Student Ambassador Program.
About The ASL Bridgify Project
ASL Bridgify is an AI-powered solution that not only serves as a dynamic tutor but also offers rapid ASL translation. It provides real-time feedback, and personalized learning experience through comprehensive AI-driven modules.
As shown in figure 1 below, the application provides modules categorized into different difficulty levels, and interactive practice sessions.
The learning modules look as follows:
The following figure shows how practice sessions look like:
The solution also provides real-time chat assistant feature as follows:
ASL Bridgify won a $25,000 grant from Academic Innovation Catalyst for ‘AI for Good’ category at the UC Berkeley’s AI Hackathon.
Intel® Tools and Technologies Used
The ASL Bridgify team trained preliminary large language models (LLMs) from scratch using pytorch-gpu environment on the Intel Tiber Developer Cloud, integrating them into a comprehensive educational platform. They used Intel® Extension for PyTorch* and Intel® Extension for TensorFlow* to train models for ASL letter and word prediction.
The team also harnessed the computational power of AI PCs provided by team Intel at the hackathon.
What’s Next?
Learn about ASL Bridgify and several other innovative AI projects developed at Berkeley AI Hackathon. Sign up to Intel Tiber Developer Cloud and try your hands on Intel’s latest accelerated hardware and optimized software for AI, HPC and edge-computing applications. Learn about Intel® AI PCs and explore AI PC development resources.
We also encourage you to dig into AI Tools and AI Frameworks optimized by Intel for accelerated AI/ML development across heterogeneous, cross-vendor architectures.
Useful Resources
- Intel Student Ambassador Program
- AI Development Resources
- AI Tools from Intel
- AI Frameworks Optimizations
- PyTorch Optimizations from Intel
- Tensorflow Optimizations from Intel
Technical Software Product Marketing Engineer, Intel