Home

Awesome

Collaborating Foundation models for Domain Generalized Semantic Segmentation

This repository contains the code for the paper: Collaborating Foundation models for Domain Generalized Semantic Segmentation.

Overview

Domain Generalized Semantic Segmentation (DGSS) deals with training a model on a labeled source domain with the aim of generalizing to unseen domains during inference. Existing DGSS methods typically effectuate robust features by means of Domain Randomization (DR). Such an approach is often limited as it can only account for style diversification and not content. In this work, we take an orthogonal approach to DGSS and propose to use an assembly of CoLlaborative FOUndation models for Domain Generalized Semantic Segmentation (CLOUDS). In detail, CLOUDS is a framework that integrates FMs of various kinds: (i) CLIP backbone for its robust feature represen- tation, (ii) text-to-image generative models to diversify the content, thereby covering various modes of the possible target distribution, and (iii) Segment Anything Model (SAM) for iteratively refining the predictions of the segmentation model. Extensive experiments show that our CLOUDS excels in adapting from synthetic to real DGSS benchmarks and under varying weather conditions, notably outperforming prior methods by 5.6% and 6.7% on averaged mIoU, respectively.

<img src="imgs/main_figure.png" width="1000"> <div style="text-align: center;"> </div>

Installation

See installation instructions.

Getting Started

See Preparing Datasets for CLOUDS.

See Getting Started with CLOUDS.

Relevant Files :

train_net.py : The training script of CLOUDS

clouds/clouds.py : This file defines the model class and its forward function, which forms the core of our model's architecture and forward pass logic

generate_txt_im.py : The script to generate a dataset using Stable Diffusion

prompt_llama70b.txt : The text file containing 100 generated prompts using Llama70b-Chat

Checkpoints & Generated dataset

We provide the following checkpoints for CLOUDS:

Citation

If you find our work useful in your research, please consider citing:

@InProceedings{Benigmim_2024_CVPR,
    author    = {Benigmim, Yasser and Roy, Subhankar and Essid, Slim and Kalogeiton, Vicky and Lathuili\`ere, St\'ephane},
    title     = {Collaborating Foundation Models for Domain Generalized Semantic Segmentation},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2024},
    pages     = {3108-3119}
}

Relevant Files :

train_net.py : The training script of CLOUDS

clouds/clouds.py : This file defines the model class and its forward function, which forms the core of our model's architecture and forward pass logic

generate_txt_im.py : The script to generate a dataset using Stable Diffusion

prompt_llama70b.txt : The text file containing 100 generated prompts using Llama70b-Chat

Acknowledgements

CLOUDS draws its foundation from the following open-source projects, and we'd like to acknowledge their authors for making their source code available :

FC-CLIP

Mask2Former

HRDA