Contrastive Pretraining with Dual Visual Encoders for Gloss-Free Sign Language Translation | Synapse