Large-scale representation learning from visually grounded untranscribed speech

Gabriel Ilharco, Yuan Zhang, Jason Baldridge

Systems that can associate images with their spoken audio captions are an important step towards visually grounded language learning. We describe a scalable method to automatically…