Florence-VL: Enhancing Vision-Language Models with Generative Vision Encoder and Depth-Breadth Fusion

TL;DR

Florence-VL introduces a novel multimodal model that integrates a generative vision encoder with depth-breadth fusion, significantly improving vision-language understanding and performance across diverse benchmarks.

Contribution

The paper presents Florence-VL, a new multimodal large language model that leverages Florence-2's versatile visual features and a novel depth-breadth fusion architecture, enhancing vision-language tasks.

Findings

Outperforms existing models on multiple vision-language benchmarks.

Enriched visual features improve vision-language alignment.

Depth-breadth fusion enhances multi-level visual feature integration.

Abstract

We present Florence-VL, a new family of multimodal large language models (MLLMs) with enriched visual representations produced by Florence-2, a generative vision foundation model. Unlike the widely used CLIP-style vision transformer trained by contrastive learning, Florence-2 can capture different levels and aspects of visual features, which are more versatile to be adapted to diverse downstream tasks. We propose a novel feature-fusion architecture and an innovative training recipe that effectively integrates Florence-2's visual features into pretrained LLMs, such as Phi 3.5 and LLama 3. In particular, we propose "depth-breath fusion (DBFusion)" to fuse the visual features extracted from different depths and under multiple prompts. Our model training is composed of end-to-end pretraining of the whole model followed by finetuning of the projection layer and the LLM, on a carefully designed recipe of diverse open-source datasets that include high-quality image captions and instruction-tuning pairs. Our quantitative analysis and visualization of Florence-VL's visual features show its advantages over popular vision encoders on vision-language alignment, where the enriched depth and breath play important roles. Florence-VL achieves significant improvements over existing state-of-the-art MLLMs across various multi-modal and vision-centric benchmarks covering general VQA, perception, hallucination, OCR, Chart, knowledge-intensive understanding, etc. To facilitate future research, our models and the complete training recipe are open-sourced. https://github.com/JiuhaiChen/Florence-VL