VALHALLA: Visual Hallucination for Machine Translation

Yi Li, Rameswar Panda, Yoon Kim, Chun-Fu (Richard) Chen, Rogerio S. Feris, David Cox, Nuno Vasconcelos; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 5216-5226


Designing better machine translation systems by considering auxiliary inputs such as images has attracted much attention in recent years. While existing methods show promising performance over the conventional text-only translation systems, they typically require paired text and image as input during inference, which limits their applicability to real-world scenarios. In this paper, we introduce a visual hallucination framework, called VALHALLA, which requires only source sentences at inference time and instead uses hallucinated visual representations for multimodal machine translation. In particular, given a source sentence an autoregressive hallucination transformer is used to predict a discrete visual representation from the input text, and the combined text and hallucinated representations are utilized to obtain the target translation. We train the hallucination transformer jointly with the translation transformer using standard backpropagation with cross-entropy losses while being guided by an additional loss that encourages consistency between predictions using either ground-truth or hallucinated visual representations. Extensive experiments on three standard translation datasets with a diverse set of language pairs demonstrate the effectiveness of our approach over both text-only baselines and state-of-the-art methods. Our codes and models will be publicly available. Project page:

Related Material

[pdf] [supp] [arXiv]
@InProceedings{Li_2022_CVPR, author = {Li, Yi and Panda, Rameswar and Kim, Yoon and Chen, Chun-Fu (Richard) and Feris, Rogerio S. and Cox, David and Vasconcelos, Nuno}, title = {VALHALLA: Visual Hallucination for Machine Translation}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, month = {June}, year = {2022}, pages = {5216-5226} }