frontend.py

import streamlit as st
import pandas as pd
import cv2
import os
from datetime import datetime
import plotly.express as px
from PIL import Image

# Set page config
st.set_page_config(
    page_title="License Plate Detection Dashboard",
    page_icon="🚗",
    layout="wide"
)

# Title
st.title("License Plate Detection Dashboard")

# Sidebar
st.sidebar.header("Dashboard Controls")

# Read the CSV file
@st.cache_data
def load_data():
    try:
        # Try to read from the original file first
        try:
            df = pd.read_csv("car_plate_data_stored.csv")
        except:
            # If that fails, try the new format file
            df = pd.read_csv("car_plate_data_new.csv")
            
        # Ensure we have the correct column names
        if 'NumberPlate' in df.columns:
            df = df.rename(columns={'NumberPlate': 'ID'})
        elif 'ImageFile' in df.columns:
            df = df.rename(columns={'ImageFile': 'ID'})
        
        # Ensure we have the right columns
        if 'NumberPlate' in df.columns:
            df = df.rename(columns={'NumberPlate': 'ID'})
        
        # Add Confidence column if it doesn't exist
        if 'Confidence' not in df.columns:
            df['Confidence'] = 'N/A'
            
        # Clean up confidence values
        df['Confidence'] = df['Confidence'].fillna('N/A')
        
        # Parse dates properly
        df['DateTime'] = pd.to_datetime(df['Date'] + ' ' + df['Time'], 
                                      format='%d-%m-%Y %H:%M:%S', 
                                      dayfirst=True)
        
        # Convert confidence values to numeric where possible
        df['Confidence'] = pd.to_numeric(df['Confidence'].replace('N/A', float('nan')), errors='coerce')
        
        # Convert date and time to datetime
        df['DateTime'] = pd.to_datetime(df['Date'] + ' ' + df['Time'])
        
        # Fill missing confidence values with None
        if 'Confidence' in df.columns:
            df['Confidence'] = df['Confidence'].fillna('N/A')
        else:
            df['Confidence'] = 'N/A'
        
        return df
    except FileNotFoundError:
        return pd.DataFrame(columns=['ID', 'Date', 'Time', 'Confidence', 'DateTime'])

df = load_data()

# Main content
col1, col2 = st.columns([2, 1])

with col1:
    st.header("Detection Statistics")
    
    # Statistics cards
    stat_col1, stat_col2, stat_col3 = st.columns(3)
    
    with stat_col1:
        st.metric("Total Detections", len(df))
    
    with stat_col2:
        if not df.empty:
            numeric_conf = pd.to_numeric(df['Confidence'].replace('N/A', float('nan')), errors='coerce')
            avg_confidence = numeric_conf.mean()
            if pd.notnull(avg_confidence):
                st.metric("Average Confidence", f"{avg_confidence:.2f}%")
            else:
                st.metric("Average Confidence", "N/A")
    
    with stat_col3:
        if not df.empty:
            today = datetime.now().date()
            today_detections = df[pd.to_datetime(df['Date']).dt.date == today].shape[0]
            st.metric("Today's Detections", today_detections)

    # Timeline chart
    if not df.empty:
        st.subheader("Detection Timeline")
        timeline_fig = px.line(
            df.groupby('DateTime').size().reset_index(name='count'),
            x='DateTime',
            y='count',
            title="Detections Over Time"
        )
        st.plotly_chart(timeline_fig, use_container_width=True)

with col2:
    st.header("Recent Detections")
    if not df.empty:
        recent = df.tail(5)
        for _, row in recent.iterrows():
            # Check if the ID is a filename (new format) or a plate number (old format)
            if row['ID'].endswith('.jpg'):
                img_path = os.path.join("detected_plates", row['ID'])
                conf_text = f" ({row['Confidence']}% confidence)" if row['Confidence'] != 'N/A' else ""
            else:
                # For old format, there might not be an image
                img_path = None
                conf_text = ""
            
            if img_path and os.path.exists(img_path):
                st.image(img_path, caption=f"Detected at {row['Time']}{conf_text}")
            else:
                st.write(f"License Plate: {row['ID']} detected at {row['Time']}{conf_text}")

# Data table
st.header("Detection Records")
if not df.empty:
    # Add filters
    date_filter = st.date_input("Filter by date", pd.to_datetime(df['Date']).min())
    confidence_filter = st.slider("Minimum confidence", 0.0, 100.0, 0.0)
    
    # Apply filters
    filtered_df = df[df['DateTime'].dt.date == date_filter]
    
    # Handle confidence filter
    if confidence_filter > 0:
        # Convert confidence to numeric, treating non-numeric values as -1
        numeric_conf = pd.to_numeric(df['Confidence'], errors='coerce').fillna(-1)
        filtered_df = filtered_df[numeric_conf >= confidence_filter]
    
    st.dataframe(
        filtered_df[['ID', 'Date', 'Time', 'Confidence']],
        use_container_width=True
    )
else:
    st.info("No detections recorded yet. Start the detection system to collect data.")

# Image gallery
st.header("Image Gallery")
if not df.empty:
    # Filter for rows that have image files
    image_df = df[df['ID'].str.endswith('.jpg', na=False)]
    if not image_df.empty:
        # Create a grid of images
        gallery = st.columns(4)
        for idx, row in image_df.iterrows():
            img_path = os.path.join("detected_plates", row['ID'])
            if os.path.exists(img_path):
                with gallery[idx % 4]:
                    conf_text = f" ({row['Confidence']}% conf.)" if row['Confidence'] != 'N/A' else ""
                    st.image(img_path, caption=f"{row['Time']}{conf_text}", use_column_width=True)
    else:
        st.info("No images available in the gallery. Only text detections are present.")