skip to Main Content
Join Us for Comet's Annual Convergence Conference on May 8-9:

SAM + Stable Diffusion for Text-to-Image Inpainting

Join the live workshop of this blog on May 9th- for free!

In this article, we’ll leverage the power of SAM, the first foundational model for computer vision, along with Stable Diffusion, a popular generative AI tool, to create a text-to-image inpainting pipeline that we’ll track in Comet. Feel free to follow along with the full code tutorial in this Colab and get the Kaggle dataset here.

Or, if you can’t wait, check out the full public project here!

Two side-by-side images. The lefthand image is the original image of a frog clinging to a tropical-colored flower with a bright, blurry background. The image on the right is the same image, only the frog has been replaced by an AI-generated koala bear with SAM + Stable Diffusion.
On the right, our original image of a frog. On the left, our output image where the frog has been replaced by a koala bear; image by author.
  Follow along with the Colab    Create a free Comet account 

What is SAM?

Earlier this year, Meta AI caused another huge stir in the computer vision community with the release of their new open-source project: the Segment Anything Model (SAM). But, what makes SAM so special? 

SAM is a prompt-able segmentation system with results that are simply stunning. It excels in zero-shot generalization to unfamiliar objects and images without the need for additional training. It’s also considered the first foundational model for computer vision, which is big news! We’ll talk a little more about foundational models next. 

SAM was trained on a massive dataset of 11 million images with 1.1 billion segmentation masks, which Meta has also released publicly. But perhaps the best way to showcase SAM’s groundbreaking capabilities is with a short demo:

 

Meta AI Segment Anything Model (SAM) demo GIF
The Segment Anything Model (SAM) is famous for correctly identifying up to hundreds of individual masks per image. It can also generate multiple valid masks for ambiguous prompts; GIF from Meta AI.

What are foundational models?

Foundation models are neural networks trained on massive unlabeled datasets to handle a wide variety of tasks. These powerful machine learning algorithms power many of the most popular Generative AI tools used today, including ChatGPT and BERT. 

Foundation models have made major strides in natural language processing, but until recently, haven’t gained much traction in computer vision applications. That’s because computer vision has struggled to find a task with semantically rich unsupervised pre-training, akin to predicting masked tokens for NLP. With SAM, Meta set out to change this.

How to use SAM

The Segment Anything Model requires no additional training, so all we need to do is provide a prompt that tells the model what to segment in a given input image. SAM accepts a variety of input prompt types, but some of the most common ones include:

Two side-by-side images. The lefthand image is the original image of a frog clinging to a tropical-colored flower with a bright, blurry background. The image on the right is the same image, only the frog has been replaced by an AI-generated koala bear with SAM + Stable Diffusion.
SAM accepts various types of input including points, bounding boxes, and text; image by author

Project Overview: GroundingDINO + SAM + Stable Diffusion

SAM doesn’t just integrate well with different input types, however. SAM’s output masks can also be used as inputs to other AI systems for even more complicated pipelines! In this tutorial, we’ll demonstrate how to use SAM in conjunction with GroundingDINO and Stable Diffusion to create a pipeline that accepts text as input to perform image inpainting and outpainting with generative AI.

A diagram of our text-to-image inpainting pipeline, including GroundingDINO for object detection, Segment Anything for segmentation masks, and Stable Diffusion for image inpainting/generative AI.
We’ll create a pipeline using GroundingDINO, Segment Anything, and Stable Diffusion to perform image inpainting with text prompts; Image by author.

To do this, we’ll be leveraging three separate models. First, we’ll use Grounding DINO to interpret our text input prompt and perform object detection for those input labels. Next, we’ll use SAM to segment the masks within those bounding box predictions. Finally, we’ll use the masks generated from SAM to isolate regions of the image for either inpainting or outpainting with Stable Diffusion. We’ll also use Comet to log the images at each step in the pipeline so we can track exactly how we got from our input image to our output image.

In the end, we should be able to provide an input image, a few input text prompts specifying what we’d like the model to do, and end up with a transformation like the one below:

 

 

Two images side-by-side. The one of the left is a picture of an orange fox on a snowy hill looking out into the distance and is captioned with word "input." The image on the right is the same image, but generative artificial intelligence has been used to replace the fox with an image of a brown bulldog. The bulldog is realistic, but the viewer can tell it isn't real. Image made with SAM + Stable Diffusion.
Our goal is to provide our pipeline with an image like the one on the left, and text prompt like the one above, and generate an output image like the one on the right; image by author.

Object detection with 🦕GroundingDINO

We’ll use four example images in this tutorial, which can be downloaded from Kaggle here. These images were all taken from Unsplash and links to the original photographers can be found at the bottom of this blog.

Our toy dataset consists of four animal images; image by author.

Once our environment is set up, we start by defining our input image and providing a text prompt that specifies which objects we want to detect. Note the format of the text prompt and make sure to separate each object with a period. We don’t have to choose from any particular categories here, so feel free to experiment with this prompt and add more categories if you’d like.

 

 

After some very simple preprocessing, we use the GroundingDINO model to predict bounding boxes for our input labels. We log these results to Comet to examine later. This way we’ll be able to see the images at each step in the pipeline, which will not only help us understand the process, but will also help us debug if anything goes wrong.

 

 

An image of two dogs sitting in front of a gray background, looking obediently at the photographer as if he is their owner. The dog on the left is a tan and white French Bulldog wearing a lavish gold and black bomber jacket. The dog on the right is a tan, brown, and white Shih Tzu wearing a flamboyant hot pink and black bomber jacket with a heavy gold chain necklace.
Our original image with the predicted bounding boxes, as visualized in Comet; image by author

We will now use these bounding box coordinates to indicate which items we would like to segment in SAM.

Masks with SAM

As mentioned, SAM can either detect all masks automatically within an image, or it can accept prompts that guide it to only detect specific masks within an image. Now that we have our bounding box predictions, we’ll use these coordinates as input prompts to SAM and plot the resulting list of binary masks: 

 

 

A grid of six black and white binary masks corresponding to the six objects detected in the original picture of two dogs. From the top left is the lefthand dog, labeled "mask 0," the righthand dog, labeled "mask 1", the shirt on the righthand dog, labeled "mask 2", the necklace on the righthand dog, marked "mask 3," the background of the picture, marked "mask 4," and finally the shirt of the lefthand dog, marked "mask 5."
The binary masks generated from SAM. Note that the number corresponds to the position of the mask in the list of masks; image by author.

Note that by default, SAM has performed instance segmentation, rather than semantic segmentation, which gives us a lot more flexibility when it comes time for inpainting. Let’s also visualize these masks within the Comet UI:

 

 

 

A GIF of our image of the two dogs, as seen in the Comet UI, with each individual segmentation mask plotted.
Examining the segmentation masks generated by SAM, as logged to Comet; GIF by author.

Finally, let’s isolate the masks we want to use for our next task: image inpainting. We’ll be replacing the dog on the right with an old man, so we’ll need the following three masks (we can grab their indices from the binary mask plot above):

 

 

Isolating part of a mask with SAM

Now, let’s say we’ve decided we want to replace the dog on the right with an old man, but just the head. If we were detecting masks with points (either interactively or programmatically), we could isolate just the dog’s face from the rest of his body using a positive and negative prompt like so:

A picture of our two dogs using a positive point prompt and a negative point prompt to indicate we just want to segment the dog's face.
The green star represents a “positive” input point and the red star represents a “negative” input point. This combination indicates to SAM that we want to segment the dog on the right, but not the body (just the face); image by author.

But since we already have our masks arrays, we’ll isolate the dog’s face using np.where. Below, we start with the mask of the dog on the right and subtract the masks for its shirt and necklace. Then we convert the array back to a PIL Image.

 

 

Image generation with Stable Diffusion

For our final step we’ll be using Stable Diffusion, a latent text-to-image deep learning model, capable of generating photo-realistic images given any text input. Specifically we’ll be using the Stable Diffusion Inpainting Pipeline, which takes as input a prompt, an image, and a binary mask image. This pipeline will generate an image from the text prompt only for the white pixels (“1”s) of the mask image.

 

 

What is inpainting?

Image inpainting refers to the process of filling-in missing data in a designated region of an image. Originally, image inpainting was used to restore damaged regions of a photo to look more like the original, but is now commonly used with masks to intentionally alter regions of an image.

Like SAM, the Stable Diffusion Inpainting Pipeline accepts both a positive and negative input prompt. Here, we instruct it to use the mask corresponding to the right dog’s face and generate “an old man with curly hair” in its place. Our negative prompt instructs the model to disclude specific objects or characteristics in the image it generates. Finally, we set the random seed so we can reproduce our results later on.

Pro tip: Stable Diffusion can be hit or miss. If you don’t like your results the first time, try adjusting the random seed and running the model again. If you still don’t like your results, try adjusting your prompts. For more on prompt engineering, read here.

 

 

An image of our two dogs but the righthand dog's face has been replaced with the face of an old man.
Our final output image with the dog’s face replaced by an old man; image by author

That was simple! Now let’s try outpainting.

What is outpainting?

Image outpainting is the process of using generative AI to extend images beyond their original borders, thereby generating parts of the image that didn’t exist before. We’ll effectively do this by masking the original background and using the same Stable Diffusion Inpainting Pipeline. 

The only difference here will be the input mask (now the background), and the input prompt. Let’s bring the dogs to Las Vegas!

 

 

A picture of our two dogs but the background has been replaced with a casino in Las Vegas. Image made with with SAM + Stable Diffusion.
Our input image after the background has been replaced by “a casino in Las Vegas”; image by author

Inpainting multiple objects with Stable Diffusion

Now let’s try segmenting more than one object in an image. In the next image we’ll ask the model to detect both the frog and the flower. We’ll then instruct the model to replace the frog with a koala bear, and replace the flower with the Empire State Building.

 

 

Two side-by-side images. The lefthand image is the original image of a frog clinging to a tropical-colored flower with a bright, blurry background. The image on the right is the same image, only the frog has been replaced by an AI-generated koala bear. Image made with with SAM + Stable Diffusion.
On the left, our original image of a frog. On the right, our output image where the frog has been replaced by a koala bear; image by author.

The model thinks the flower includes the frog, but we can work around that by subtracting the frog mask and then converting the new mask to a PIL image.

Three binary masks in a row showing the result of subtracting one mask from another.
From left to right: the original flower mask, minus the frog mask, equals our new, corrected flower mask; image by author.

Once we’ve separated the flower, let’s replace it with the Empire State Building:

 

 

Three images in a row. The first is a tropical frog, the second replaces the frog with a koala bear using generative AI, and the third rreplaces the flower with a skyscraper (the Empire State Building) using generative AI. Image made with with SAM + Stable Diffusion.
On the left, our original image. In the center, we’ve used SAM + Stable Diffusion to replace the frog with a koala bear, and on the right we’ve also replaced the flower with a skyscraper; image by author.

Our model isn’t perfect; it looks like our koala may have a fifth leg, and there’s still some remnants of frog on the skyscraper, but generally, our pipeline performed pretty well!

Defining the background for Stable Diffusion

Sometimes our object detector, GroundingDINO won’t detect the background. But we can still easily perform outpainting!

To create a background mask when one isn’t detected, we can just take the inverse of the object mask. If multiple objects are in the image, we would just add these masks together, and then take the inverse of this sum.

 

 

We can then follow the same process as in the previous examples.

Viewing our SAM + Stable Diffusion results in Comet

As you can probably imagine, keeping track of which input images, prompts, masks, and random seeds were used to create which output images can get confusing, fast! That’s why we logged all of our images to Comet as we went. 

Let’s head on over to the Comet UI now and take a look at each of our input images and the resulting output images after inpainting and outpainting:

A grid of input and output images as represented in the Comet UI. Images made with with SAM + Stable Diffusion.
We create a clean, simple dashboard to track our input images and the final output images. Which is your favorite? Image by author.

That’s a nice, clean dashboard, but sometimes we want to get a deeper understanding of how we went from point A to point B. Or, maybe, something has gone wrong and we need to take a  deeper look at each step of the process to debug. For this, we’ll check our custom Debugging dashboard:

Our debugging dashboard for our text-to-image inpainting pipeline, as seen in the Comet UI
Our dashboard showing each step of each image in our pipeline; GIF by author.

We can also take a closer look at each step of an individual experiment:

A screenshot of all steps of the image pipeline, as visualized in the Comet UI.
We also create a second dashboard detailing each step of the process for each image. This is helpful for debugging if something goes wrong in our pipeline; GIF by author.

Tracking our prompts with Comet

We’ll also want to make sure to keep track of how we created each output so we can reproduce any of the results later on. Maybe we’ve run different versions of the same prompt multiple times. Or maybe we’ve tried different random seeds and want to pick our favorite result. By logging our prompts to Comet’s Data Panel, we can easily retrieve all the relevant information to recreate any of our image outputs.

A GIF of Comet's Data Panel used to track text prompts and random seeds used to create Stable Diffusion image outputs.
Tracking and organizing our prompt and seed information with Comet’s Data Panels; GIF by author

Now that you’re an inpainting pro, try out the pipeline on your own images!

Three pictures; on the left, a picture of a seal underwater looking straight at the camera. In the center, the seal has been replaced by a bright purple octopus. On the right, the background has been replaced by a comfortable bed. Image made with with SAM + Stable Diffusion.
Try out the pipeline on your own images!

Conclusion

Thanks for making it all the way to the end, and I hope you found this SAM + Stable Diffusion tutorial helpful! For questions, comments, or feedback, feel free to drop a note in the comments below. Happy coding!

Frequently Asked Questions (FAQs)

Where can I find the full code for this tutorial?

Find the full code tutorial in this Colab here

Where can I find the images used in this tutorial?

Download the dataset on Kaggle here.

Is Segment Anything open source?

Yes, the Segment Anything model and dataset are both open source! Find the code for the Segment Anything Model here and download the full Segment Anything dataset here.

How many images were used to train SAM?

The Segment Anything dataset consists of 100 million images with 1.1 billion segmentation masks. It’s roughly 400x larger than the next largest segmentation mask dataset and 6x larger than OpenImages V5.

Is the Segment Anything Model a foundational model?

Yes, the Segment Anything model is considered the first foundational model for computer vision.

Is there a Segment Anything paper?

Yes, there is a Segment Anything paper available for free on arxiv.org here.

Is there a Segment Anything demo?

There is a free interactive Segment Anything demo available online here

Is there a Segment Anything API?

There is not currently a Segment Anything API, but the code is open source and available on GitHub.

Is there a Stable Diffusion repo?

Yes, there is a Stable Diffusion GitHub repository.

Is there a Stable Diffusion API?

Yes, there is a Stable Diffusion and Dreambooth API. Learn more here.

Where can I create a Comet account?

Sign up for a free Comet account here

Image Credits

The sample images used in this tutorial were all downloaded originally from Unsplash:

Join the live workshop of this blog on May 9th- for free!

Abby Morgan

Machine Learning Marketing Engineer @ Comet
Back To Top