gagaolala/instructcv

Codebase for "InstructCV: Instruction-Tuned Text-to-Image Diffusion Models as Vision Generalists"

InstructCV: Instruction-Tuned Text-to-Image Diffusion Models as Vision Generalists

Yulu Gan, Sungwoo Park, Alexander Schubert, Anthony Philippakis and Ahmed Alaa

Paper | HuggingFace 🤗 Demo

🌟 Official PyTorch implementation of InstructCV. The master branch works with PyTorch 1.5+.

InstructCVDemo.mp4

Overview

Recent advances in generative diffusion models have enabled text-controlled synthesis of realistic and diverse images with impressive quality. Despite these remarkable advances, the application of text-to-image generative models in computer vision for standard visual recognition tasks remains limited. The current de facto approach for these tasks is to design model architectures and loss functions that are tailored to the task at hand. In this project, we develop a unified language interface for computer vision tasks that abstracts away task specific design choices and enables task execution by following natural language instructions. Our approach involves casting multiple computer vision tasks as text-to-image generation problems. Here, the text represents an instruction describing the task, and the resulting image is a visually-encoded task output. To train our model, we pool commonly-used computer vision datasets covering a range of tasks, including segmentation, object detection, depth estimation, and classification. We then use a large language model to paraphrase prompt templates that convey the specific tasks to be conducted on each image, and through this process, we create a multi-modal and multi-task training dataset comprising input and output images along with annotated instructions. Following the InstructPix2Pix architecture, we apply instruction-tuning to a text-to-image diffusion model using our constructed dataset, steering its functionality from a generative model to an instruction-guided multi-task vision learner.

Set up the environments

Install dependencies by running:

#Step0. Set up the env.
conda env create -f environment.yaml
conda activate lvi
#Step1 (optional) . You could ignore this step if you do not run the baselines.
## install tensorflow : https://www.tensorflow.org/install/pip
pip install git+https://github.com/philferriere/cocoapi.git#subdirectory=PythonAPI 
pip install -U openmim
mim install mmcv-full
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -v -e .

Get Started

See Preparing Datasets for InstructCV.

See Getting Started with InstructCV for detailed instructions on training and inference with InstructCV.

InstructCV-RP checkpoint

	Depth Estimation RMSE⬇		Sementic Segmentation mIoU⬆		Classification Acc⬆		Object Detection mAP⬆		Download
	NYUv2	SUNRGB-D	ADE-20K	VOC	Oxford-Pets	ImageNet-sub	COCO	VOC
InstructCV-RP	0.297	0.279	47.235	52.125	82.135	74.665	48.500	61.700	checkpoint

Demo

• Integrated into Huggingface Spaces 🤗 using Gradio. Try out the web demo:

• Run the demo on Google Colab:

The pre-trained model for Stable Diffusion is subject to its original license terms from Stable Diffusion.

Acknowledgement

This codebase is largely based on CompVis/stable_diffusion and Instruct Pix2Pix.

Citation

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

@article{gan2023instructcv,
  title={InstructCV: Instruction-Tuned Text-to-Image Diffusion Models as Vision Generalists},
  author={Gan, Yulu and Park, Sungwoo and Schubert, Alexander and Philippakis, Anthony and Alaa, Ahmed},
  journal={arXiv preprint arXiv:2310.00390},
  year={2023}
}