classifier.py

import pandas as pd
import numpy as np
from sklearn.datasets import fetch_openml
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from PIL import Image
import PIL.ImageOps 

X= np.load("image.npz")['arr_0']
y = pd.read_csv("labels.csv")["labels"]

print(pd.Series(y).value_counts())

classes = ['A','B','C','D','E','F','G','H','I','J','K','L','M','N','O','P','Q','R','S','T','U','V','W','X','Y','Z']

nclasses= len(classes)

X_train,X_test,y_train,y_test = train_test_split(X,y,train_size = 3500,test_size=500,random_state=42)

X_train_scaled = X_train/255
X_test_scaled = X_test/255

clf = LogisticRegression(solver="saga",multi_class="multinomial").fit(X_train_scaled,y_train)

def get_prediction(image):
    im_pil = Image.open(image)
    image_bw =im_pil.convert("L")
    image_bw_resized=image_bw.resize((28,28),Image.ANTIALIAS) 
    pixel_filter = 20
    min_pixel = np.percentile(image_bw_resized,pixel_filter)
    image_bw_resized_scaled = np.clip(image_bw_resized-min_pixel,0,255)
    max_pixel = np.max(image_bw_resized)
    image_bw_resized_scaled=np.asarray(image_bw_resized_scaled)/max_pixel
    test_sample = np.array(image_bw_resized_scaled).reshape(1,660)
    test_pred = clf.predict(test_sample)
    return test_pred[0]