VL-BERT: Pre-training of Generic Visual-Linguistic Representations
This addresses the need for unified models in visual-linguistic tasks, offering a pre-trained solution that benefits applications like visual commonsense reasoning and question answering, though it is incremental as it builds on existing Transformer architectures.
The paper tackles the problem of creating a generic visual-linguistic representation by introducing VL-BERT, which extends Transformer to handle both visual and linguistic inputs, and demonstrates its effectiveness by achieving first place on the VCR benchmark with a single model.
We introduce a new pre-trainable generic representation for visual-linguistic tasks, called Visual-Linguistic BERT (VL-BERT for short). VL-BERT adopts the simple yet powerful Transformer model as the backbone, and extends it to take both visual and linguistic embedded features as input. In it, each element of the input is either of a word from the input sentence, or a region-of-interest (RoI) from the input image. It is designed to fit for most of the visual-linguistic downstream tasks. To better exploit the generic representation, we pre-train VL-BERT on the massive-scale Conceptual Captions dataset, together with text-only corpus. Extensive empirical analysis demonstrates that the pre-training procedure can better align the visual-linguistic clues and benefit the downstream tasks, such as visual commonsense reasoning, visual question answering and referring expression comprehension. It is worth noting that VL-BERT achieved the first place of single model on the leaderboard of the VCR benchmark. Code is released at \url{https://github.com/jackroos/VL-BERT}.