FlaskML is a web application that leverages the Flask framework and the Deep Dream algorithm to provide an interactive and user-friendly experience for exploring the concept of neural network-based image generation. The application allows users to upload their own images and generate dream-like versions of them using the Deep Dream algorithm. 🎨 🌌 It also provides a brief explanation of the concepts behind Deep Dream, including the concept of activation maximization through Gradient Ascent.
Deep Dream is a computer vision algorithm that generates highly abstract and dream-like images by amplifying the activations of the neurons in a deep convolutional neural network trained on image classification tasks. For a more detailed explanation visit the original Deep Dream blogpost.
The application was built to utilize pre-trained deep learning models for the Deep Dream algorithm, implemented by Aleksa Gordic. The user interface was designed for simplicity and ease of use, allowing users to quickly upload images and view the dreamified results.
It is not anywhere near perfect but I am working (and learning) to improve it while balancing the daily responsibilities I have.
git clone https://github.com/LuisMongeB/flaskml.git
- Open Anaconda Prompt and navigate into project directory
cd path_to_repo
- Create your own environment using Python version 3.9.15 like this
conda create --name flaskml python=3.9.15
and runpip install -r requirementstxt
- Activate your environment
conda activate YOUR_ENV_NAME
after creating your own environment. - Last we need to create the database models. Open terminal and run
python
. Then, runfrom app import app, db
followed bywith app.app_context(): db.create_all()
and then pressCTR-C
to exit the Python interpreter. - In the terminal:
flask run
and you're done!
For any issues or comments, reach out please!
- Python 3.9.15
- PyTorch 1.13.1
- Torchvision 0.14.1
- Flask (with Bootstrap and Jinja2 templating)
- Sqlite3 (with SQLAlchemy) and DB Browser for SQLite
Contributions, issues, and feature requests are welcome! Give a ⭐️ if you like this project!