Large Hole Image Inpainting


AI model preview image
The MAT model is a neural network-based model that can be used for large hole image inpainting. It is able to fill in large missing regions in an image by predicting the missing content based on the surrounding context. The model uses a transformer architecture along with a mask-aware mechanism to effectively handle large holes in the image. It achieves state-of-the-art performance on several benchmark datasets for image inpainting tasks.

Use cases

The MAT model for large hole image inpainting has a wide range of potential use cases for various industries. In the field of digital editing and design, this model can be used to seamlessly remove unwanted objects or structures from images, allowing for more aesthetically pleasing compositions. It can also be employed in the restoration of old or damaged photographs, intelligently filling in missing areas and preserving the integrity of the original image. In the field of surveillance and security, this model can be used for enhancing and reconstructing images with obscured or missing data, aiding in forensic investigations. Additionally, in the realm of virtual reality and gaming, the MAT model can be utilized to generate realistic and immersive environments by efficiently inpainting large areas that are yet to be rendered or captured. In terms of practical applications, this model can be integrated into photo editing software, restoration tools, or even image and video compression algorithms to improve the quality and visual integrity of the final output.



Cost per run
Avg run time
Nvidia T4 GPU

Creator Models

No other models by this creator

Similar Models

Try it!

You can use this area to play around with demo applications that incorporate the Large Hole Image Inpainting model. These demos are maintained and hosted externally by third-party creators. If you see an error, message me on Twitter.

Currently, there are no demos available for this model.


Summary of this model and related resources.

Model NameLarge Hole Image Inpainting
MAT: Mask-Aware Transformer for Large Hole Image Inpainting
Model LinkView on Replicate
API SpecView on Replicate
Github LinkView on Github
Paper LinkView on Arxiv


How popular is this model, by number of runs? How popular is the creator, by the sum of all their runs?

Model Rank
Creator Rank


How much does it cost to run this model? How long, on average, does it take to complete a run?

Cost per Run$0.0011
Prediction HardwareNvidia T4 GPU
Average Completion Time2 seconds