app.py

import numpy as np
import pandas as pd
from ultralytics import YOLO
import streamlit as st
import cv2
import base64
import time
import shutil
import os
from PIL import Image
import base64


st.set_page_config(layout="wide",initial_sidebar_state="expanded",
                   page_icon='🔎',page_title='Poth-Hole Detector')


def get_video_base64(video_path):
    with open(video_path, "rb") as file:
        video_bytes = file.read()
        base64_encoded = base64.b64encode(video_bytes).decode("utf-8")
        return base64_encoded

video_path = "deep1.mp4"
video_base64 = get_video_base64(video_path)

video_html = f"""
	<style>
	#myVideo {{
		position: fixed;
		right: 0;
		bottom: 0;
		min-width: 100%; 
		min-height: 100%;
	}}
	.content {{
		position: fixed;
		bottom: 0;
		background: rgba(0, 0, 0, 0.5);
		color: #f1f1f1;
		width: 100%;
		padding: 20px;
	}}

	</style>

	<video autoplay loop muted id="myVideo">
		<source type="video/mp4" src="data:video/mp4;base64,{video_base64}">
	</video>
"""

st.markdown(video_html, unsafe_allow_html=True)

# Define custom style for the glowing text
glowing_text_style = '''
    <style>
        .glowing-text {
            font-family: 'Arial Black', sans-serif;
            font-size: 48px;
            text-align: center;
            animation: glowing 2s infinite;
        }
        
        @keyframes glowing {
            0% { color: #FF9933; } /* Saffron color */
            25% { color: #FFFFFF; } /* White color */
            50% { color: #128807; } /* Green color */
            75% { color: #0000FF; } /* Blue color */
            100% { color: #FF9933; } /* Saffron color */
        }
    </style>
'''

# Display the glowing text using st.markdown
st.markdown(glowing_text_style, unsafe_allow_html=True)
st.markdown(f'<p class="glowing-text">🕳️ PothHole Detector 🕳️</p>', unsafe_allow_html=True)

def upload():
    image=None
    image_filename=None
    initial_image = st.camera_input('Take a picture')
    original_image = initial_image
    temp_path = None
    if initial_image is not None:
        image_filename = f"{int(time.time())}.jpg"
        bytes_data = initial_image.getvalue()
        image = cv2.imdecode(np.frombuffer(bytes_data, np.uint8), cv2.IMREAD_COLOR)

    return image, original_image,image_filename

def process_line(line, image_np,counter):
    # Process a single line from the labels.txt file
    bresults = line.split()
    if len(bresults) >=5:
        names={0:'POTH_HOLE'}
        xc, yc, nw, nh = map(float, bresults[1:5])
        h, w = image_np.shape[0], image_np.shape[1]

        xc *= w
        yc *= h
        nw *= w
        nh *= h
        top_left = (int(xc - nw / 2), int(yc - nh / 2))
        bottom_right = (int(xc + nw / 2), int(yc + nh / 2))

        # Draw bounding box
        cv2.rectangle(image_np, top_left, bottom_right, (4, 29, 255), 3, cv2.LINE_4)

        # Draw label text
        #label = names[int(bresults[0])]
        label = f'{names[int(bresults[0])]}-{counter}'
        text_size = cv2.getTextSize(label, cv2.FONT_HERSHEY_SIMPLEX, 2, 3)[0]
        text_width, text_height = text_size
        text_x = (top_left[0] + bottom_right[0] - text_width) // 2 + 100
        text_y = top_left[1] - 10
        cv2.putText(image_np, label, (text_x, text_y), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 0), 2)
        

 
sidebar_option = st.sidebar.radio("Select an option", ("Take picture for prediction", "Upload file"))

def main():
    
    
    
    
   
    if sidebar_option == "Take picture for prediction":
        if st.checkbox('Take a picture for prediction'):
    
        
            image, original_image,image_filename= upload()
            if original_image is not None and original_image is not None and len(image_filename)!=0 and st.button('Prediction'):  # Check if original_image is not None
                st.info('Wait for the results...!')
                #image1=cv2.imread(image)
                
                counter=1
                names={0:'POTH_HOLE'
                    }
               
                model=YOLO('best.pt')
                result = model.predict(image,save=True,save_txt=True)
                txt_files_exist = any(filename.endswith('.txt') for filename in os.listdir('runs/detect/predict/labels'))
                if txt_files_exist:
                    lis=open('runs/detect/predict/labels/image0.txt','r').readlines()
                    for line in lis:
                        process_line(line, image_np,counter)
                        counter+=1
                    with st.spinner('Wait for the results...!'):
                        time.sleep(5)
                    
                    st.image(image,use_column_width=True)
                    st.balloons()
                    
                    try:
                        if os.path.exists('runs'):
                            shutil.rmtree('runs')
                            st.session_state.original_image = None  # Clear the original_image variable
                           
                    except Exception as e:
                        st.error(f"An error occurred: {e}")
                else:
                    st.warning('⚠️Please check your image')
                    try:
                        if os.path.exists('runs'):
                            shutil.rmtree('runs')
                            st.session_state.original_image = None  # Clear the original_image variable
                           
                    except Exception as e:
                        st.error(f"An error occurred: {e}")
    elif sidebar_option == "Upload file":  
                  
        
        fileimage=st.file_uploader('Upload the file for detection 📁',type=['jpg'])
       
        if st.button('Predict'):
            if fileimage is None:
                st.warning('🛎️We are using default image 📷')
                counter=1
                default_image_path='df.jpg'
                pic = Image.open(default_image_path)
                image_np = np.array(pic)
                names={0:'POTH_HOLE'
                        }
                mod1=YOLO('best.pt')
                mod1.predict(image_np,save=True,save_txt=True)
                txt_files_exist = any(filename.endswith('.txt') for filename in os.listdir('runs/detect/predict/labels'))
                if txt_files_exist:
                    lis=open('runs/detect/predict/labels/image0.txt','r').readlines()
                    with st.spinner('Wait for the results...!'):
                        time.sleep(5)
    
                                    
                    for line in lis:
                        process_line(line, image_np,counter)
                        counter+=1
                    col1,col2=st.columns(2)
                    with col1:
                                
                        st.info('Original Image!')
                        st.image(default_image_path,use_column_width=True)
                    with col2:
                        st.info('Detected Image!')
                        st.image(image_np,use_column_width=True)
                    st.balloons()
                        
                    try:
                        if os.path.exists('runs'):
                            shutil.rmtree('runs')
                            st.session_state.original_image = None  # Clear the original_image variable
                                
                    except Exception as e:
                        st.error(f"An error occurred: {e}")
                    
            else:
                    
                if fileimage is not None and not fileimage.name.endswith('.jpg'):
                    st.warning('⚠️ Please upload only JPG images.')
                else:
                    st.info('Wait for the results...!')
                    counter=1
                    pic=Image.open(fileimage)
                    image_np = np.array(pic)
                    names={0:'POTH_HOLE'
                        }
                    mod1=YOLO('best.pt')
                    mod1.predict(image_np,save=True,save_txt=True)
                    txt_files_exist = any(filename.endswith('.txt') for filename in os.listdir('runs/detect/predict/labels'))
                    if txt_files_exist:
                        lis=open('runs/detect/predict/labels/image0.txt','r').readlines()
                        with st.spinner('Wait for the results...!'):
                            time.sleep(5)
    
                                    
                        for line in lis:
                            process_line(line, image_np,counter)
                            counter+=1
                        col1,col2=st.columns(2)
                        with col1:
                                
                            st.info('Original Image!')
                            st.image(fileimage,use_column_width=True)
                        with col2:
                            st.info('Detected Image!')
                            st.image(image_np,use_column_width=True)
                        st.balloons()
                        
                        try:
                            if os.path.exists('runs'):
                                shutil.rmtree('runs')
                                st.session_state.original_image = None  # Clear the original_image variable
                                
                        except Exception as e:
                            st.error(f"An error occurred: {e}")
                    else:
                        st.warning('⚠️Please check your image')
                        try:
                            if os.path.exists('runs'):
                                shutil.rmtree('runs')
                                st.session_state.original_image = None  # Clear the original_image variable
                            
                        except Exception as e:
                            st.error(f"An error occurred: {e}")
                    

        
        
            

   
if __name__ == '__main__':
    
   
    main()