Do cues in a video help in handling rare words in a machine translation system under a low-resource setting?
Abstract: Highlights•Enhance the performance of a machine translation system under a low resource setting.•Build the first synthetic English-Hindi multimodal dataset with video modality.•Identify the encoding technique that captures the key information in a video.•Extensive analysis of text-only NMT and attention-based video-guided MMT systems.•Demonstrate the significance of video cues in a machine translation system.
Loading