@article{wang2020minivlm, author = {Wang, Jianfeng and Hu, Xiaowei and Zhang, Pengchuan and Li, Xiujun and Wang, Lijuan and Zhang, Lei and Gao, Jianfeng and Liu, Zicheng}, title = {MiniVLM: A Smaller and Faster Vision-Language Model}, year = {2020}, month = {December}, abstract = {Recent vision-language (VL) studies have shown remarkable progress by learning generic representations from massive image-text pairs with transformer models and then fine-tuning on downstream VL tasks. While existing research has been focused on achieving high accuracy with large pre-trained models, building a lightweight model is of great value in practice but is less explored. In this paper, we propose a smaller and faster VL model, MiniVLM, which can be finetuned with good performance on various downstream tasks like its larger counterpart. MiniVLM consists of two modules, a vision feature extractor and a transformer-based vision-language fusion module. We design a Two-stage Efficient feature Extractor (TEE), inspired by the one-stage EfficientDet network, to significantly reduce the time cost of visual feature extraction by 95%, compared to a baseline model. We adopt the MiniLM structure to reduce the computation cost of the transformer module after comparing different compact BERT models. In addition, we improve the MiniVLM pre-training by adding 7M Open Images data, which are pseudo-labeled by a state-of-the-art captioning model. We also pre-train with high-quality image tags obtained from a strong tagging model to enhance cross-modality alignment. The large models are used offline without adding any overhead in fine-tuning and inference. With the above design choices, our MiniVLM reduces the model size by 73% and the inference time cost by 94% while being able to retain 94−97% of the accuracy on multiple VL tasks. We hope that MiniVLM helps ease the use of the state-of-the-art VL research for on-the-edge applications.}, url = {http://approjects.co.za/?big=en-us/research/publication/minivlm-a-smaller-and-faster-vision-language-model/}, journal = {arXiv preprint arXiv:2012.06946}, }