تشخیص مهره شطرنج

import cv2
import numpy as np
import pandas as pd
import glob
import os
import re
from tqdm import tqdm
import matplotlib as mpl
import matplotlib.pyplot as plt

CHESS_BOARD_DIR = glob.glob('train/*.*')
CHESS_BOARD = []

def normalizeBoardName(boardName):
    mine_code=[]
    longName=boardName.split('-')
    for word in longName:
        for letter in word:
            if letter.isnumeric():
                for i in range(int(letter)):
                    mine_code.append(str(0))
            else:
                mine_code.append(letter)
                
    return mine_code

df = pd.DataFrame()
points=[5,19,  20,34,  35,49,  50,64,  64,78,  79,93,  94,108,  109,123]
my_counter=0
for idx, path in enumerate(CHESS_BOARD_DIR):
    my_counter_local=0
    if idx==1000:
        break
    
    image = cv2.imread(path)
    image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    CHESS_BOARD = [path[6:-4], image]
    PEACE_CODES=normalizeBoardName(CHESS_BOARD[0])
    #--------------------------------------------
    
    for j in range(0,len(points) , 2):# columns
        
        for i in range (0,len(points) , 2):# rows
            for board in CHESS_BOARD:
                img_small = CHESS_BOARD[1]
                img_small=img_small[points[j]:points[j+1],points[i]:points[i+1]] #first col , then row
                
#         cv2.imshow('s', img_small)
#         cv2.waitKey(0)
#         cv2.destroyAllWindows()
#         print(type(img_small))


            img_small = img_small.reshape(14*14)
            img_small=np.append(img_small,PEACE_CODES[my_counter_local])
            df[my_counter]=img_small
            my_counter+=1
            my_counter_local+=1
        
# df=df.T
# df_1=df[df[196]!='0']
# df_1.shape
# df_1.head()
# df_1.to_csv('all_images.csv', sep=',')

df_1=pd.read_csv('all_images.csv')
df_1= df_1.drop('Unnamed: 0',axis=1)

X = df_1.drop('196', axis=1)

y = df_1['196']

y.unique()

from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
label = le.fit_transform(df_1['196'])
label

X_train, X_test, \
y_train, y_test =\
X[:14000], X[14000:], y[:14000], y[14000:]

from sklearn.linear_model import SGDClassifier
sgd_clf = SGDClassifier(random_state=42)
sgd_clf.fit(X_train, y_train)

sgd_clf.predict(pd.DataFrame(X.loc[0]).T)