ImageSense is a versatile prototype application designed for image analysis using a pre-trained Machine Learning model, with built-in support for future translation into various languages. Please note that the current prototype utilizes the ViT + GPT2 model for basic image descriptions, and it is not well-trained for production environments. It is intended for study and design purposes.
-
Demo: Explore a functional demo version of the application hosted on my personal server. Check it out here!(Not available anymore!) -
Frontend:
- Developed with React.js.
- Enables users to intuitively select images, crop specific parts, and send them for further processing using a Machine Learning model.
-
Backend:
- Implemented in Python using Flask.
- Consumes a pre-trained machine learning model.
- Offers flexibility to replace the model with other image processing machine learning models.
-
Design:
- Features a straightforward design.
- Utilizes Heroicons and TailwindCSS for styling components.
- Mostly uses CSS flexbox.
-
Language Support:
- Built with future translation in mind, allowing seamless integration of additional languages. (Note: There is a known bug in the translation function that triggers a warning but still works.)
-
Deployment:
- Frontend hosted on a personal server.
- Backend hosted on Google Cloud Platform Compute Engine, adaptable to any available Virtual Machine.
- The Flask app runs using Supervisor and Gunicorn. Configuration details are available in the deployment environment and are not provided here for simplicity.
- Nginx was used as a reverse proxy to route external traffic to the Gunicorn server.
- HTTPS traffic was enabled using a self-signed certificate for added security.
All Rights Reserved.
- Model: ViT + GPT2 Model by Ankur Kumar.
- Icons: [Heroicons](https://heroicons.com