streamlit_app.py

import os
import streamlit as st
import pandas as pd
import joblib
import numpy as np
import cv2
from PIL import Image
from rembg import remove
from skimage.feature import graycomatrix, graycoprops

# Loading the model
model = open("models/knn_coconut_min_max.pkl", "rb")
knn = joblib.load(model)

# Create title
st.title("Coconut Quality Classification App")

# Upload image 
uploaded_file = st.file_uploader("Upload an image", type=["jpg", "jpeg", "png"])

# Create the 'temps/' directory if it doesn't exist
os.makedirs('temps/', exist_ok=True)

# check if the file is uploaded
if uploaded_file is not None:
    # Read the image
    image = Image.open(uploaded_file)
    # Set the image path
    temp_image_path = 'temps/temp_image.png'
    # Remove the background of the image
    output = remove(image)
    # Crop the image to the contents within the bounding box
    output = output.crop(output.getbbox())
    # Resize image to 224x224 pixels
    output = output.resize((224, 224))
    # Save the processed image
    output.save(temp_image_path)
    
    # Display the processed image
    st.image(output, caption='Uploaded Image', use_column_width=True)
    
    # Ekstraksi Fitur Warna dan Bentuk
    rataR = []
    rataG = []
    rataB = []
    rataH = []
    rataS = []
    rataV = []
    ratagray = []
    stand = []

    luas = []
    keliling = []

    contrast = []
    dissimilarity = []
    homogeneity = []
    energy = []
    correlation = []
    
    gbr_read = cv2.imread(temp_image_path)
    gbr_rgb = cv2.cvtColor(gbr_read, cv2.COLOR_BGR2RGB)
    (R, G, B) = cv2.split(gbr_rgb)
    meanR = np.mean(R)
    rataR.append(meanR)
    meanG = np.mean(G)
    rataG.append(meanG)
    meanB = np.mean(B)
    rataB.append(meanB)
    
    gbr_hsv = cv2.cvtColor(gbr_read, cv2.COLOR_BGR2HSV)
    H = gbr_hsv[:, :, 0]
    S = gbr_hsv[:, :, 1]
    V = gbr_hsv[:, :, 2]
    meanH = np.mean(H)
    rataH.append(meanH)
    meanS = np.mean(S)
    rataS.append(meanS)
    meanV = np.mean(V)
    rataV.append(meanV)

    gbr_gray = cv2.cvtColor(gbr_read, cv2.COLOR_BGR2GRAY)
    rata = np.mean(gbr_gray)
    ratagray.append(rata)
    standar = np.std(gbr_gray)
    stand.append(standar)  
    
    _, thresh = cv2.threshold(gbr_gray, 127, 255, 0)
    contours, _ = cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    
    if len(contours) > 0:
        contour = max(contours, key=cv2.contourArea)
        area = cv2.contourArea(contour)
        perimeter_val = cv2.arcLength(contour, True)
    else:
        area = 0
        perimeter_val = 0
    
    luas.append(area)
    keliling.append(perimeter_val)

    # Ekstraksi Fitur Tekstur (GLCM)
    glcm = graycomatrix(gbr_gray, [1], [0], 256, symmetric=True, normed=True)
    contrast_val = graycoprops(glcm, 'contrast')[0, 0]
    dissimilarity_val = graycoprops(glcm, 'dissimilarity')[0, 0]
    homogeneity_val = graycoprops(glcm, 'homogeneity')[0, 0]
    energy_val = graycoprops(glcm, 'energy')[0, 0]
    correlation_val = graycoprops(glcm, 'correlation')[0, 0]
    
    contrast.append(contrast_val)
    dissimilarity.append(dissimilarity_val)
    homogeneity.append(homogeneity_val)
    energy.append(energy_val)
    correlation.append(correlation_val)
    
    # Memasukkan data ke dalam dataframe
    data1 = pd.DataFrame(rataR, columns=['Mean-R'])
    data2 = pd.DataFrame(rataG, columns=['Mean-G'])
    data3 = pd.DataFrame(rataB, columns=['Mean-B'])
    data4 = pd.DataFrame(rataH, columns=['Mean-H'])
    data5 = pd.DataFrame(rataV, columns=['Mean-V'])
    data6 = pd.DataFrame(rataS, columns=['Mean-s'])
    data7 = pd.DataFrame(ratagray, columns=['Mean-Gray'])
    data8 = pd.DataFrame(stand, columns=['Standar-Deviasi'])
    data9 = pd.DataFrame(luas, columns=['Luas'])
    data10 = pd.DataFrame(keliling, columns=['Keliling'])
    data11 = pd.DataFrame(contrast, columns=['Contrast'])
    data12 = pd.DataFrame(dissimilarity, columns=['Dissimilarity'])
    data13 = pd.DataFrame(homogeneity, columns=['Homogeneity'])
    data14 = pd.DataFrame(energy, columns=['Energy'])
    data15 = pd.DataFrame(correlation, columns=['Correlation'])
    
    # Memasukkan data ke dalam list
    listdata = [data1, data2, data3, data4, data5, data6, data7, data8, data9, 
                data10, data11, data12, data13, data14, data15]
    # Menggabungkan data dari list
    x = pd.concat(listdata, axis=1, ignore_index=True)
    
    # Menampilkan data
    st.write(x) 
    
    # Prediksi Kelapa Standar atau Tidak Standar
    y_prediksi = knn.predict(x)
    
    if y_prediksi == 0:
        st.write("<p style='text-align: center;'><center>Kelapa Standar</center></p>", unsafe_allow_html=True)
    elif y_prediksi == 1:
        st.write("<p style='text-align: center;'><center>Kelapa Tidak Standar</center></p>", unsafe_allow_html=True)
    else:
        st.write("<p style='text-align: center;'><center>error</center></p>", unsafe_allow_html=True)