https://github.com/sofiia-chorna/clothes-segmentation-project

Web-service that allows users to segment clothes on the images using elements of machine learning.
https://github.com/sofiia-chorna/clothes-segmentation-project

matplotlib neural-network python pytorch streamlit-webapp

Last synced: 9 days ago
JSON representation

Web-service that allows users to segment clothes on the images using elements of machine learning.

• Host: GitHub
• URL: https://github.com/sofiia-chorna/clothes-segmentation-project
• Owner: sofiia-chorna
• Created: 2022-11-12T05:29:45.000Z (almost 2 years ago)
• Default Branch: develop
• Last Pushed: 2022-12-11T19:41:55.000Z (almost 2 years ago)
• Last Synced: 2024-10-14T08:14:36.218Z (24 days ago)
• Topics: matplotlib, neural-network, python, pytorch, streamlit-webapp
• Language: Python
• Homepage:
• Size: 4.23 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Project Clothes Segmentation

## Summary

This project involved using Deep Convolutional Neural network to create a machine learning application that could classify clothes based on images. The trained model is going to be deployed in an interactive website to allow users to identify their own pictures.

## Technologies used: 

```

* Python

* Pytorch

* Matprotlib

* Streamlit

```

## Results

 ![](result.png)

## Installation

1. Clone project`s [repo](https://github.com/sofiia-chorna/clothes-segmentation):

    ```

    git clone https://github.com/sofiia-chorna/clothes-segmentation.git

    ```

2. Install the required packages.

    ```

    pip install -r requirements.txt

    ```

   

3. In the command line (terminal) go to the ```src``` folder:

    ```

    cd /* path to src folder */

    ```

   

4. To run the application from the command line (terminal) in the project folder, run:

    ```

    streamlit run app.py

    ```

5. View the application in your default browser by navigating to the following URL:

    ``` 

   http://localhost:8501

   ```

## Future direcitons

I have several ideas to improve this project:

* Add explanations for how the CNN works depending on user selection of dropbox

* If predicted confidence is under some threshold, say something about not being sure about the prediction

* Potentially have a stacked model where the first model predicts if the image is a clothes or not - if not, do something funny to the user for trying to trick me