Uni-EDEN: Universal Encoder-Decoder Network by Multi-Granular Vision-Language Pre-training

TL;DR

Uni-EDEN is a versatile Transformer-based model pre-trained on multi-granular vision-language tasks, enabling effective perception and generation in various vision-language applications with strong generalizability.

Contribution

The paper introduces Uni-EDEN, a universal encoder-decoder network that jointly learns multi-modal perception and generation through multi-granular pre-training tasks, unlike existing single-encoder models.

Findings

Effective across multiple vision-language tasks

Strong generalization after fine-tuning

Outperforms previous models in perception and generation

Abstract

Vision-language pre-training has been an emerging and fast-developing research topic, which transfers multi-modal knowledge from rich-resource pre-training task to limited-resource downstream tasks. Unlike existing works that predominantly learn a single generic encoder, we present a pre-trainable Universal Encoder-DEcoder Network (Uni-EDEN) to facilitate both vision-language perception (e.g., visual question answering) and generation (e.g., image captioning). Uni-EDEN is a two-stream Transformer based structure, consisting of three modules: object and sentence encoders that separately learns the representations of each modality, and sentence decoder that enables both multi-modal reasoning and sentence generation via inter-modal interaction. Considering that the linguistic representations of each image can span different granularities in this hierarchy including, from simple to comprehensive, individual label, a phrase, and a natural sentence, we pre-train Uni-EDEN through multi-granular vision-language proxy tasks: Masked Object Classification (MOC), Masked Region Phrase Generation (MRPG), Image-Sentence Matching (ISM), and Masked Sentence Generation (MSG). In this way, Uni-EDEN is endowed with the power of both multi-modal representation extraction and language modeling. Extensive experiments demonstrate the compelling generalizability of Uni-EDEN by fine-tuning it to four vision-language perception and generation downstream tasks.