Fashion Recommendation System

Project Description

This project is a fashion recommendation system designed to suggest similar fashion products based on a user's uploaded image. The system leverages a pre-trained VGG16 model to extract features from the uploaded image and compare it to a dataset of fashion products to find the most similar items.

Dataset

The dataset used in this project is the "Fashion Product Images (Small)" dataset from Kaggle. It contains information about various fashion products, including their images and style details.

Downloading the Dataset

You can download the dataset from Kaggle using the following link: Fashion Product Images (Small)

To download the dataset, follow these steps:

Go to the Kaggle dataset page.
Click on the "Download" button to download the dataset files.

The dataset contains a CSV file (styles.csv) with the following columns:

id: Unique identifier for each product
gender: Gender category of the product (e.g., Men, Women)
masterCategory: Main category of the product (e.g., Apparel)
subCategory: Sub-category of the product (e.g., Topwear)
articleType: Specific type of article (e.g., T-Shirts)
baseColour: Primary color of the product
season: Season associated with the product
year: Year the product was released
usage: Usage category (e.g., Casual)
productDisplayName: Display name of the product

The images corresponding to these products are named with their id followed by .jpg.

File Structure

app.py: Main application script that implements the recommendation system.
styles.csv: Dataset file containing details about fashion products.
fashion-products-recommendation-system.ipynb: Jupyter notebook containing additional analysis and code for the recommendation system.

Requirements

To run this project, you need the following packages installed:

numpy
pandas
tensorflow
keras
matplotlib
cv2
streamlit
PIL
joblib

You can install these packages using pip:

pip install numpy pandas tensorflow keras matplotlib opencv-python-headless streamlit pillow joblib

Running the Project

Prepare the Dataset:

Ensure that styles.csv is located in the root directory of the project. Also, ensure that the images are stored in a directory named fashion_small/images/.
Run the Streamlit Application:

Execute the following command to start the Streamlit application:
```
streamlit run app.py
```
Upload an Image:

Open the Streamlit application in your web browser. You will be prompted to upload an image. Choose a fashion product image (in jpg, jpeg, or png format).
View Recommendations:

Once the image is uploaded, the system will process the image and display the top 5 recommended images that are similar to the uploaded image.

Code Overview

app.py

This script contains the main logic for the recommendation system. It defines a fashion_recommendations class that includes methods for loading the model, processing the image, calculating similarity, and generating recommendations. The main function sets up the Streamlit interface for uploading an image and displaying the recommendations.

Helper Functions

get_styles_df: Loads the dataset and prepares it by adding an image column.
load_model: Loads a pre-trained VGG16 model with ImageNet weights and prepares it for feature extraction.
predict: Preprocesses the input image and uses the model to extract features.
get_similarity: Calculates similarity between the input image features and the dataset features.
normalize_sim: Normalizes the similarity scores.
get_recommendations: Retrieves the top 5 most similar products based on similarity scores.
print_recommendations: Saves and displays the recommended images.

Preprocess Image

preprocess_image: Preprocesses the uploaded image to the required format for the model.

Main Function

main: Sets up the Streamlit interface, handles file upload, and displays recommendations.

Conclusion

This fashion recommendation system provides a simple yet effective way to find similar fashion products based on an uploaded image. By leveraging a pre-trained VGG16 model and a dataset of fashion products, the system can help users discover new fashion items that match their preferences.

Geo-y20/Fashion-Recommendation-System