Lexicon-Free Fingerspelling Recognition from Video: Data, Models, and Signer Adaptation
This addresses the challenge of fingerspelling recognition for ASL users and researchers, with incremental improvements through data collection and model adaptation.
The paper tackles the problem of recognizing continuous fingerspelling in American Sign Language from video, achieving up to 92% letter accuracy in signer-dependent settings and 83% with adaptation for multi-signer scenarios.
We study the problem of recognizing video sequences of fingerspelled letters in American Sign Language (ASL). Fingerspelling comprises a significant but relatively understudied part of ASL. Recognizing fingerspelling is challenging for a number of reasons: It involves quick, small motions that are often highly coarticulated; it exhibits significant variation between signers; and there has been a dearth of continuous fingerspelling data collected. In this work we collect and annotate a new data set of continuous fingerspelling videos, compare several types of recognizers, and explore the problem of signer variation. Our best-performing models are segmental (semi-Markov) conditional random fields using deep neural network-based features. In the signer-dependent setting, our recognizers achieve up to about 92% letter accuracy. The multi-signer setting is much more challenging, but with neural network adaptation we achieve up to 83% letter accuracies in this setting.