Florence-VL: Enhancing Vision-Language Models with Generative Vision Encoder and Depth-Breadth Fusion
About
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
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Object Hallucination Evaluation | POPE | -- | 2019 | |
| Visual Question Answering | TextVQA | Accuracy74.2 | 1453 | |
| Text-based Visual Question Answering | TextVQA | Accuracy69.1 | 962 | |
| Multimodal Understanding | MMBench | -- | 847 | |
| Science Question Answering | ScienceQA | Accuracy84.6 | 791 | |
| Multimodal Understanding | MM-Vet | MM-Vet Score56.3 | 631 | |
| Visual Question Answering | ChartQA | Accuracy74.7 | 519 | |
| Multimodal Understanding | SEED-Bench | Accuracy70.6 | 516 | |
| Mathematical Reasoning | MathVista | Score55.5 | 474 | |
| Multimodal Understanding | MMStar | -- | 407 |