Text to image generation and Image Captioning Android, iOS, Desktop and Web app using Compose Multiplatform with Clean Architecture.
- Moko for ViewModels
- Voyager for navigation
- Ktor for HTTP communication
- Sql Delight as local database for Images History
- Koin for dependency injection
- Calf for adaptive UI Elements
- Typist-CMP for text typing animation
App is following the MVVM Clean Architecture approach with buildSrc for dependencies management.
- App is currently using HuggingFace's free inference API Directly but you can also use the backend linked below. Instructions for changing to custom backend are given in data source in code. Get your token from Hugging Face and replace in data layer to start using the app. (You don't need to add "Bearer" before it, it is automatically handeld by bearerAuth function of Ktor)
- Django App hosted on AWS EC2 (which further uses HuggingFace's free inference api for custom trained model)
- HuggingFace Hosted Model
You can either use the above backend and deploy it or you can directly use HuggingFace's free inference API for generating images (Currently Implemented Method).
If you face any error mentioning Sql Delight or sqlite3, refer to this issue
Note: HuggingFace's free inference API is initialised on demand so at first request you will receive invalid response and after a while, when system is initialised you need to try again to start getting the actual results.
