Florence: A New Foundation Model for Computer Vision [article]

Lu Yuan and Dongdong Chen and Yi-Ling Chen and Noel Codella and Xiyang Dai and Jianfeng Gao and Houdong Hu and Xuedong Huang and Boxin Li and Chunyuan Li and Ce Liu and Mengchen Liu and Zicheng Liu and Yumao Lu and Yu Shi and Lijuan Wang and Jianfeng Wang and Bin Xiao and Zhen Xiao and Jianwei Yang and Michael Zeng and Luowei Zhou and Pengchuan Zhang
2021 arXiv   pre-print
Automated visual understanding of our diverse and open world demands computer vision models to generalize well with minimal customization for specific tasks, similar to human vision. Computer vision foundation models, which are trained on diverse, large-scale dataset and can be adapted to a wide range of downstream tasks, are critical for this mission to solve real-world computer vision applications. While existing vision foundation models such as CLIP, ALIGN, and Wu Dao 2.0 focus mainly on
more » ... ing images and textual representations to a cross-modal shared representation, we introduce a new computer vision foundation model, Florence, to expand the representations from coarse (scene) to fine (object), from static (images) to dynamic (videos), and from RGB to multiple modalities (caption, depth). By incorporating universal visual-language representations from Web-scale image-text data, our Florence model can be easily adapted for various computer vision tasks, such as classification, retrieval, object detection, VQA, image caption, video retrieval and action recognition. Moreover, Florence demonstrates outstanding performance in many types of transfer learning: fully sampled fine-tuning, linear probing, few-shot transfer and zero-shot transfer for novel images and objects. All of these properties are critical for our vision foundation model to serve general purpose vision tasks. Florence achieves new state-of-the-art results in majority of 44 representative benchmarks, e.g., ImageNet-1K zero-shot classification with top-1 accuracy of 83.74 and the top-5 accuracy of 97.18, 62.4 mAP on COCO fine tuning, 80.36 on VQA, and 87.8 on Kinetics-600.
arXiv:2111.11432v1 fatcat:qpq2twmmgrapxb6whq3f4iss6u