Grounded-Segment-Anything

We plan to create a very interesting demo by combining Grounding DINO and Segment Anything which aims to detect and segment Anything with text inputs! And we will continue to improve it and create more interesting demos based on this foundation.

We are very willing to help everyone share and promote new projects based on Segment-Anything, Please checkout here for more amazing demos and works in the community: Highlight Extension Projects. You can submit a new issue (with project tag) or a new pull request to add new project's links.

🍄 Why Building this Project?

The core idea behind this project is to combine the strengths of different models in order to build a very powerful pipeline for solving complex problems. And it's worth mentioning that this is a workflow for combining strong expert models, where all parts can be used separately or in combination, and can be replaced with any similar but different models (like replacing Grounding DINO with GLIP or other detectors / replacing Stable-Diffusion with ControlNet or GLIGEN/ Combining with ChatGPT).

🍇 Updates

• Chech out the great re-implementation of DragGAN by Shanghai AI Lab.
• 2023/05/23: Support Image-Referring-Segment, Audio-Referring-Segment and Text-Referring-Segment in ImageBind-SAM.
• 2023/05/16: Release ImageBind-SAM simple demo which aims to segment with different modalities.
• 2023/05/15: Release LaMa and RePaint demo, thanks for nice tips by Tao Yu.
• 2023/05/14: Release PaintByExample demo with SAM.
• 2023/05/11: We decide to share more interesting demo in playground and we've already tested the DeepFloyd for image generation and style transfering and share some notes about using IF.
• 2023/05/05: Release a simpler code for automatic labeling (combined with Tag2Text model): please see automatic_label_simple_demo.py
• 2023/05/03: Checkout the Automated Dataset Annotation and Evaluation with GroundingDINO and SAM which is an amazing tutorial on automatic labeling! Thanks a lot for Piotr Skalski and Robotflow!

Table of Contents

• Grounded-Segment-Anything
  • Preliminary Works
  • Highlighted Projects
• Installation
  • Install with Docker
  • Install locally
• Grounded-SAM Playground
  • Step-by-Step Notebook Demo
  • GroundingDINO: Detect Everything with Text Prompt
  • Grounded-SAM: Detect and Segment Everything with Text Prompt
  • Grounded-SAM with Inpainting: Detect, Segment and Generate Everything with Text Prompt
  • Grounded-SAM and Inpaint Gradio APP
  • Grounded-SAM with Tag2Text for Automatic Labeling
  • Grounded-SAM with BLIP & ChatGPT for Automatic Labeling
  • Grounded-SAM with Whisper: Detect and Segment Anything with Audio
  • Grounded-SAM ChatBot with Visual ChatGPT
  • Grounded-SAM with OSX for 3D Whole-Body Mesh Recovery
  • Grounded-SAM with VISAM for Tracking and Segment Anything
  • Interactive Fashion-Edit Playground: Click for Segmentation And Editing
  • Interactive Human-face Editing Playground: Click And Editing Human Face
  • 3D Box Via Segment Anything
  • Playground: More Interesting and Imaginative Demos with Grounded-SAM
    • DeepFloyd: Image Generation with Text Prompt
    • PaintByExample: Exemplar-based Image Editing with Diffusion Models
    • LaMa: Resolution-robust Large Mask Inpainting with Fourier Convolutions
    • RePaint: Inpainting using Denoising Diffusion Probabilistic Models
    • ImageBind with SAM: Segment with Different Modalities

Preliminary Works

Here we provide some background knowledge that you may need to know before trying the demos.

Title	Intro	Description	Links
Segment-Anything		A strong foundation model aims to segment everything in an image, which needs prompts (as boxes/points/text) to generate masks	[Github] [Page] [Demo]
Grounding DINO		A strong zero-shot detector which is capable of to generate high quality boxes and labels with free-form text.	[Github] [Demo]
OSX		A strong and efficient one-stage motion capture method to generate high quality 3D human mesh from monucular image. OSX also releases a large-scale upper-body dataset UBody for a more accurate reconstrution in the upper-body scene.	[Github] [Page] [Video] [Data]
Stable-Diffusion		A super powerful open-source latent text-to-image diffusion model	[Github] [Page]
BLIP		A wonderful language-vision model for image understanding.	[GitHub]
Visual ChatGPT		A wonderful tool that connects ChatGPT and a series of Visual Foundation Models to enable sending and receiving images during chatting.	[Github] [Demo]
Tag2Text		An efficient and controllable vision-language model which can simultaneously output superior image captioning and image tagging.	[Github] [Demo]
VoxelNeXt		A clean, simple, and fully-sparse 3D object detector, which predicts objects directly upon sparse voxel features.	[Github]

Highlighted Projects

Here we provide some impressive works you may find interesting:

Title	Description	Links
SEEM: Segment Everything Everywhere All at Once	A powerful promptable segmentation model supports segmenting with various types of prompts (text, point, scribble, referring image, etc.) and any combination of prompts.	[Github] [Demo]
OpenSeeD	A simple framework for open-vocabulary segmentation and detection which supports interactive segmentation with box input to generate mask	[Github]
LLaVA	Visual instruction tuning with GPT-4	[Github] [Page] [Demo] [Data] [Model]

We also list some awesome segment-anything extension projects here you may find interesting:

• Computer Vision in the Wild (CVinW) Readings for those who are interested in open-set tasks in computer vision.
• Zero-Shot Anomaly Detection by Yunkang Cao
• EditAnything: ControlNet + StableDiffusion based on the SAM segmentation mask by Shanghua Gao and Pan Zhou
• IEA: Image Editing Anything by Zhengcong Fei
• SAM-MMRorate: Combining Rotated Object Detector and SAM by Qingyun Li and Xue Yang
• Awesome-Anything by Gongfan Fang
• Prompt-Segment-Anything by Rockey
• WebUI for Segment-Anything and Grounded-SAM by Chengsong Zhang
• Inpainting Anything: Inpaint Anything with SAM + Inpainting models by Tao Yu
• Grounded Segment Anything From Objects to Parts: Combining Segment-Anything with VLPart & GLIP & Visual ChatGPT by Peize Sun and Shoufa Chen
• Narapi-SAM: Integration of Segment Anything into Narapi (A nice viewer for SAM) by MIC-DKFZ
• Grounded Segment Anything Colab by camenduru
• Optical Character Recognition with Segment Anything by Zhenhua Yang
• Transform Image into Unique Paragraph with ChatGPT, BLIP2, OFA, GRIT, Segment Anything, ControlNet by showlab
• Lang-Segment-Anything: Another awesome demo for combining GroundingDINO with Segment-Anything by Luca Medeiros
• 🥳 🚀 Playground: Integrate SAM and OpenMMLab!
• 3D-object via Segment Anything by Yukang Chen
• Image2Paragraph: Transform Image Into Unique Paragraph by Show Lab
• Zero-shot Scene Graph Generate with Grounded-SAM by JackWhite-rwx
• CLIP Surgery for Better Explainability with Enhancement in Open-Vocabulary Tasks by Eli-YiLi
• Panoptic-Segment-Anything: Zero-shot panoptic segmentation using SAM by segments-ai
• Caption-Anything: Generates Descriptive Captions for Any Object within an Image by Teng Wang
• Segment-Anything-3D: Transferring Segmentation Information of 2D Images to 3D Space by Yunhan Yang
• Expediting SAM without Fine-tuning by Weicong Liang and Yuhui Yuan
• Semantic Segment Anything: Providing Rich Sementic Category Annotations for SAM by Jiaqi Chen and Zeyu Yang and Li Zhang
• Enhance Everything: Combining SAM with Image Restoration and Enhancement Tasks by Xin Li

Installation

The code requires python>=3.8, as well as pytorch>=1.7 and torchvision>=0.8. Please follow the instructions here to install both PyTorch and TorchVision dependencies. Installing both PyTorch and TorchVision with CUDA support is strongly recommended.

Install with Docker

Open one terminal:

make build-image

make run

That's it.

If you would like to allow visualization across docker container, open another terminal and type:

xhost +

Install without Docker

You should set the environment variable manually as follows if you want to build a local GPU environment for Grounded-SAM:

export AM_I_DOCKER=False
export BUILD_WITH_CUDA=True
export CUDA_HOME=/path/to/cuda-11.3/

Install Segment Anything:

python -m pip install -e segment_anything

Install Grounding DINO:

python -m pip install -e GroundingDINO

Install diffusers:

pip install --upgrade diffusers[torch]

Install osx:

git submodule update --init --recursive
cd grounded-sam-osx && bash install.sh

Install Tag2Text:

git submodule update --init --recursive
cd Tag2Text && pip install -r requirements.txt

The following optional dependencies are necessary for mask post-processing, saving masks in COCO format, the example notebooks, and exporting the model in ONNX format. jupyter is also required to run the example notebooks.

pip install opencv-python pycocotools matplotlib onnxruntime onnx ipykernel

More details can be found in install segment anything and install GroundingDINO and install OSX

Grounded-SAM Playground

Let's start exploring our Grounding-SAM Playground and we will release more interesting demos in the future, stay tuned!

📖 Step-by-Step Notebook Demo

Here we list some notebook demo provided in this project:

• grounded_sam.ipynb
• grounded_sam_colab_demo.ipynb
• grounded_sam_3d_box.ipynb

🏃‍♂️ GroundingDINO: Detect Everything with Text Prompt

🍇 [arXiv Paper]   🌹[Try the Colab Demo]   🌻 [Try Huggingface Demo]   🍄 [Automated Dataset Annotation and Evaluation]

Here's the step-by-step tutorial on running GroundingDINO demo:

Step 1: Download the pretrained weights

cd Grounded-Segment-Anything

# download the pretrained groundingdino-swin-tiny model
wget https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth

Step 2: Running the demo

python grounding_dino_demo.py

Running with Python (same as demo but you can run it anywhere after installing GroundingDINO)

from groundingdino.util.inference import load_model, load_image, predict, annotate
import cv2

model = load_model("GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py", "./groundingdino_swint_ogc.pth")
IMAGE_PATH = "assets/demo1.jpg"
TEXT_PROMPT = "bear."
BOX_TRESHOLD = 0.35
TEXT_TRESHOLD = 0.25

image_source, image = load_image(IMAGE_PATH)

boxes, logits, phrases = predict(
    model=model,
    image=image,
    caption=TEXT_PROMPT,
    box_threshold=BOX_TRESHOLD,
    text_threshold=TEXT_TRESHOLD
)

annotated_frame = annotate(image_source=image_source, boxes=boxes, logits=logits, phrases=phrases)
cv2.imwrite("annotated_image.jpg", annotated_frame)

Tips

• If you want to detect multiple objects in one sentence with Grounding DINO, we suggest seperating each name with . . An example: cat . dog . chair .

Step 3: Check the annotated image

The annotated image will be saved as ./annotated_image.jpg.

Text Prompt	Demo Image	Annotated Image
Bear.
Horse. Clouds. Grasses. Sky. Hill

🏃‍♂️ Grounded-SAM: Detect and Segment Everything with Text Prompt

Here's the step-by-step tutorial on running Grounded-SAM demo:

Step 1: Download the pretrained weights

cd Grounded-Segment-Anything

wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth
wget https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth

We provide two versions of Grounded-SAM demo here:

• grounded_sam_demo.py: our original implementation for Grounded-SAM.
• grounded_sam_simple_demo.py our updated more elegant version for Grounded-SAM.

Step 2: Running original grounded-sam demo

export CUDA_VISIBLE_DEVICES=0
python grounded_sam_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/demo1.jpg \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --text_prompt "bear" \
  --device "cuda"

The annotated results will be saved in ./outputs as follows

Input Image	Annotated Image	Generated Mask

Step 3: Runing the updated grounded-sam demo (optional) Note that this demo is almost same as the original demo, but with more elegant code.

python grounded_sam_simple_demo.py

The annotated results will be saved as ./groundingdino_annotated_image.jpg and ./grounded_sam_annotated_image.jpg

Text Prompt	Input Image	GroundingDINO Annotated Image	Grounded-SAM Annotated Image
The running dog
Horse. Clouds. Grasses. Sky. Hill

⛷️ Grounded-SAM with Inpainting: Detect, Segment and Generate Everything with Text Prompt

Step 1: Download the pretrained weights

cd Grounded-Segment-Anything

wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth
wget https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth

Step 2: Running grounded-sam inpainting demo

CUDA_VISIBLE_DEVICES=0
python grounded_sam_inpainting_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/inpaint_demo.jpg \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --det_prompt "bench" \
  --inpaint_prompt "A sofa, high quality, detailed" \
  --device "cuda"

The annotated and inpaint image will be saved in ./outputs

Step 3: Check the results

Input Image	Det Prompt	Annotated Image	Inpaint Prompt	Inpaint Image
	Bench		A sofa, high quality, detailed

🏌️ Grounded-SAM and Inpaint Gradio APP

We support 6 tasks in the local Gradio APP：

1. scribble: Segmentation is achieved through Segment Anything and mouse click interaction (you need to click on the object with the mouse, no need to specify the prompt).
2. automask: Segment the entire image at once through Segment Anything (no need to specify a prompt).
3. det: Realize detection through Grounding DINO and text interaction (text prompt needs to be specified).
4. seg: Realize text interaction by combining Grounding DINO and Segment Anything to realize detection + segmentation (need to specify text prompt).
5. inpainting: By combining Grounding DINO + Segment Anything + Stable Diffusion to achieve text exchange and replace the target object (need to specify text prompt and inpaint prompt) .
6. automatic: By combining BLIP + Grounding DINO + Segment Anything to achieve non-interactive detection + segmentation (no need to specify prompt).

python gradio_app.py

• The gradio_app visualization as follow:

🏷️ Grounded-SAM with Tag2Text for Automatic Labeling

Tag2Text achieves superior image tag recognition ability of 3,429 commonly human-used categories. It is seamlessly linked to generate pseudo labels automatically as follows:

1. Use Tag2Text to generate tags.
2. Use Grounded-Segment-Anything to generate the boxes and masks.

Step 1: Init submodule and download the pretrained checkpoint

• Init submodule:

cd Grounded-Segment-Anything
git submodule init
git submodule update

• Download pretrained weights for GroundingDINO, SAM and Tag2Text:

wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth
wget https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth

cd Tag2Text
wget https://huggingface.co/spaces/xinyu1205/Tag2Text/resolve/main/tag2text_swin_14m.pth

Step 2: Runing the demo

export CUDA_VISIBLE_DEVICES=0
python automatic_label_tag2text_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --tag2text_checkpoint ./Tag2Text/tag2text_swin_14m.pth \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/demo9.jpg \
  --output_dir "outputs" \
  --box_threshold 0.25 \
  --text_threshold 0.2 \
  --iou_threshold 0.5 \
  --device "cuda"

• Tag2Text also provides powerful captioning capabilities, and the process with captions can refer to BLIP.
• The pseudo labels and model prediction visualization will be saved in output_dir as follows (right figure):

🤖 Grounded-SAM with BLIP for Automatic Labeling

It is easy to generate pseudo labels automatically as follows:

1. Use BLIP (or other caption models) to generate a caption.
2. Extract tags from the caption. We use ChatGPT to handle the potential complicated sentences.
3. Use Grounded-Segment-Anything to generate the boxes and masks.

• Run Demo

export OPENAI_API_KEY=your_openai_key
export OPENAI_API_BASE=https://closeai.deno.dev/v1
export CUDA_VISIBLE_DEVICES=0
python automatic_label_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/demo3.jpg \
  --output_dir "outputs" \
  --openai_key $OPENAI_API_KEY \
  --box_threshold 0.25 \
  --text_threshold 0.2 \
  --iou_threshold 0.5 \
  --device "cuda"

• When you don't have a paid Account for ChatGPT is also possible to use NLTK instead. Just don't include the openai_key Parameter when starting the Demo.
  • The Script will automatically download the necessary NLTK Data.
• The pseudo labels and model prediction visualization will be saved in output_dir as follows:

😮 Grounded-SAM with Whisper: Detect and Segment Anything with Audio

Detect and segment anything with speech!

Install Whisper

pip install -U openai-whisper

See the whisper official page if you have other questions for the installation.

Run Voice-to-Label Demo

Optional: Download the demo audio file

wget https://huggingface.co/ShilongLiu/GroundingDINO/resolve/main/demo_audio.mp3

export CUDA_VISIBLE_DEVICES=0
python grounded_sam_whisper_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/demo4.jpg \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --speech_file "demo_audio.mp3" \
  --device "cuda"

Run Voice-to-inpaint Demo

You can enable chatgpt to help you automatically detect the object and inpainting order with --enable_chatgpt.

Or you can specify the object you want to inpaint [stored in args.det_speech_file] and the text you want to inpaint with [stored in args.inpaint_speech_file].

export OPENAI_API_KEY=your_openai_key
export OPENAI_API_BASE=https://closeai.deno.dev/v1
# Example: enable chatgpt
export CUDA_VISIBLE_DEVICES=0
python grounded_sam_whisper_inpainting_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/inpaint_demo.jpg \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --prompt_speech_file assets/acoustics/prompt_speech_file.mp3 \
  --enable_chatgpt \
  --openai_key $OPENAI_API_KEY\
  --device "cuda"

# Example: without chatgpt
export CUDA_VISIBLE_DEVICES=0
python grounded_sam_whisper_inpainting_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --input_image assets/inpaint_demo.jpg \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --det_speech_file "assets/acoustics/det_voice.mp3" \
  --inpaint_speech_file "assets/acoustics/inpaint_voice.mp3" \
  --device "cuda"

💬 Grounded-SAM ChatBot Demo

chatbot.mp4

Following Visual ChatGPT, we add a ChatBot for our project. Currently, it supports:

1. "Descripe the image."
2. "Detect the dog (and the cat) in the image."
3. "Segment anything in the image."
4. "Segment the dog (and the cat) in the image."
5. "Help me label the image."
6. "Replace the dog with a cat in the image."

To use the ChatBot:

• Install whisper if you want to use audio as input.
• Set the default model setting in the tool Grounded_dino_sam_inpainting.
• Run Demo

export OPENAI_API_KEY=your_openai_key
export OPENAI_API_BASE=https://closeai.deno.dev/v1
export CUDA_VISIBLE_DEVICES=0
python chatbot.py 

🕺 Run Grounded-Segment-Anything + OSX Demo

• Download the checkpoint osx_l_wo_decoder.pth.tar from here for OSX:

• Download the human model files and place it into grounded-sam-osx/utils/human_model_files following the instruction of OSX.

• Run Demo

export CUDA_VISIBLE_DEVICES=0
python grounded_sam_osx_demo.py \
  --config GroundingDINO/groundingdino/config/GroundingDINO_SwinT_OGC.py \
  --grounded_checkpoint groundingdino_swint_ogc.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --osx_checkpoint osx_l_wo_decoder.pth.tar \
  --input_image assets/osx/grounded_sam_osx_demo.png \
  --output_dir "outputs" \
  --box_threshold 0.3 \
  --text_threshold 0.25 \
  --text_prompt "humans, chairs" \
  --device "cuda"

• The model prediction visualization will be saved in output_dir as follow:

• We also support promptable 3D whole-body mesh recovery. For example, you can track someone with with a text prompt and estimate his 3D pose and shape :

A person with pink clothes

A man with a sunglasses

🕺 Run Grounded-Segment-Anything + VISAM Demo

• Download the checkpoint motrv2_dancetrack.pth from here for MOTRv2:

• See the more thing if you have other questions for the installation.

• Run Demo

export CUDA_VISIBLE_DEVICES=0
python grounded_sam_visam.py \
  --meta_arch motr \
  --dataset_file e2e_dance \
  --with_box_refine \
  --query_interaction_layer QIMv2 \
  --num_queries 10 \
  --det_db det_db_motrv2.json \
  --use_checkpoint \
  --mot_path your_data_path \
  --resume motrv2_dancetrack.pth \
  --sam_checkpoint sam_vit_h_4b8939.pth \
  --video_path DanceTrack/test/dancetrack0003 

||

👯 Interactive Editing

• Release the interactive fashion-edit playground in here. Run in the notebook, just click for annotating points for further segmentation. Enjoy it!

  

• Release human-face-edit branch here. We'll keep updating this branch with more interesting features. Here are some examples:

  

📷 3D-Box via Segment Anything

We extend the scope to 3D world by combining Segment Anything and VoxelNeXt. When we provide a prompt (e.g., a point / box), the result is not only 2D segmentation mask, but also 3D boxes. Please check voxelnext_3d_box for more details.

💘 Acknowledgements

• Segment Anything
• Grounding DINO

Contributors

Our project wouldn't be possible without the contributions of these amazing people! Thank you all for making this project better.

Citation

If you find this project helpful for your research, please consider citing the following BibTeX entry.

@article{kirillov2023segany,
  title={Segment Anything}, 
  author={Kirillov, Alexander and Mintun, Eric and Ravi, Nikhila and Mao, Hanzi and Rolland, Chloe and Gustafson, Laura and Xiao, Tete and Whitehead, Spencer and Berg, Alexander C. and Lo, Wan-Yen and Doll{\'a}r, Piotr and Girshick, Ross},
  journal={arXiv:2304.02643},
  year={2023}
}

@article{liu2023grounding,
  title={Grounding dino: Marrying dino with grounded pre-training for open-set object detection},
  author={Liu, Shilong and Zeng, Zhaoyang and Ren, Tianhe and Li, Feng and Zhang, Hao and Yang, Jie and Li, Chunyuan and Yang, Jianwei and Su, Hang and Zhu, Jun and others},
  journal={arXiv preprint arXiv:2303.05499},
  year={2023}
}