SimpleAI_Image

A package for processing and storing image vectors in PostgreSQL.

Installation

pip install SimpleAI_Image

from SimpleAI_Image import DatabaseHandler, DataProcessor
from sklearn.datasets import fetch_openml
from tensorflow.keras.applications.vgg16 import preprocess_input as vgg_preprocess_input

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sqlalchemy import text

# Define the database URL and table name
db_url = 'postgresql+psycopg2://Username:Password@localhost:5432/ThisISATEST'
db_handler = DatabaseHandler(db_url, 'vector_data', 512)

# Instantiate DataProcessor with VGG16 model
data_processor = DataProcessor(db_handler, model_name='VGG16', preprocess_func=vgg_preprocess_input, image_size=(32, 32))

# Load the example dataset (MNIST)
mnist = fetch_openml('mnist_784', version=1)
X = mnist.data[:500]  # Limit to 500 instances for testing
y = mnist.target[:500].astype(int)  # Ensure targets are integers

# Process data and store in database
X_embedded, y = data_processor.process_data(X, y)

# Fetch and preprocess data for visualization
query = text("SELECT * FROM vector_data")
X_embedded, y = data_processor.fetch_and_preprocess_data(query)

# Visualize data
data_processor.visualize_data(X_embedded, y)