Gaston Rios

I'm a researcher in Computer Science at Universidad Nacional de La Plata focusing on Sign Language Recognition and Translation. My work primarily involves developing deep learning models to bridge communication gaps between deaf and hearing communities.

Currently, my research centers on creating efficient, low-parameter approaches for Sign Language Recognition and Translation. I've been working on pose-based models and innovative architectures like the ConvAtt Network, which combines 1D convolutions with transformer layers to achieve competitive results while using fewer parameters. I collaborate closely with researchers like Franco Ronchetti, F. Quiroga, and W. Hasperué to tackle challenges in handshape recognition and gloss-free sign language translation.

My recent work has focused on addressing the critical challenge of limited data availability in sign language datasets. I've been exploring various techniques including data augmentation, regularization, and few-shot learning approaches to improve model performance with small datasets. Through this research, we're working towards making sign language translation technology more accessible and practical for real-world applications.