Skip to content

arongsnuna/Dear_Image

Repository files navigation

Dear Image: Integrated Image Editing Service Using Visprog

imddddage

Overview

Dear Image is an innovative image editing service that simplifies the editing process for users who find traditional software like Photoshop overwhelming due to its complexity and subscription costs. This service combines various image editing functionalities into a single platform, making it accessible for everyone without requiring a login.

Background

Many users are unable to utilize powerful image editing tools due to the steep learning curve and costs associated with them. Dear Image addresses this gap by providing a comprehensive suite of image editing tools accessible through a simple chat interface.

Key Features:

  • Unified Functionality: All editing tools are integrated into one platform.
  • No Login Required: Users can access services without creating an account.
  • Command-Based Interface: Users can edit images using straightforward commands.

Features

Modular Image Editing

Dear Image uses compositional visual reasoning to allow users to edit images through modular commands. The system is powered by a large language model (LLM) that generates programs based on user commands.

Key Functionalities:

  • Blur: Apply a blur effect to selected objects in an image.
  • Replace: Swap one object in an image for another.
  • GQA (Grouped Query Attention): Answer questions about the contents of an image.
  • Color Pop: Highlight selected objects by turning the rest of the image to grayscale.
  • Remove Background: Eliminate the background from images entirely.

Expected Benefits

  • User-Friendly: Dear Imagesimplifies image editing tasks, reducing user burdens associated with image processing.
  • Accessibility: The web-based platform allows easy access without login requirements.
  • AI-Powered Efficiency: Leveraging AI enhances the accuracy and speed of image modifications.

Conclusion

Dear Image aims to revolutionize the way users interact with image editing by providing a straightforward, chat-based approach that accommodates both novices and experienced users alike.

Contributors

GitHub Repositories

Demo

Check out the demo video showcasing the functionalities of Visprog here.

References

Gupta, T., & Kembhavi, A. (2022). Visual Programming: Compositional visual reasoning without training. arXiv, 2211.11559.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages