app.py

import streamlit as st
import cv2
import os
import numpy as np
from datetime import datetime
import matplotlib.pyplot as plt
import time
from services.detection_service import detect_faults_solar, detect_faults_windmill
from services.anomaly_service import track_anomalies, predict_anomaly
from models.solar_model import load_solar_model
from models.windmill_model import load_windmill_model
from config.settings import VIDEO_FOLDER
from PIL import Image
import io

# Set page config as the first Streamlit command
st.set_page_config(page_title="Thermal Anomaly Monitoring Dashboard", layout="wide")

# Custom CSS for styling
st.markdown(
    """
    <style>
    .main-header {
        text-align: center;
        font-size: 24px;
        font-weight: bold;
        color: #333;
    }
    .status {
        text-align: center;
        font-size: 16px;
        color: #333;
        margin-bottom: 20px;
    }
    .section-title {
        font-size: 16px;
        font-weight: bold;
        color: #333;
        text-transform: uppercase;
        margin-bottom: 10px;
    }
    .section-box {
        border: 1px solid #4A90E2;
        padding: 10px;
        border-radius: 5px;
        margin-bottom: 20px;
    }
    .log-entry {
        font-size: 14px;
        color: #333;
        margin-bottom: 5px;
    }
    .metrics-text {
        font-size: 14px;
        color: #333;
        margin-bottom: 5px;
    }
    .snapshot-img {
        max-width: 100%;
        height: auto;
        margin-bottom: 10px;
    }
    </style>
    """,
    unsafe_allow_html=True
)

# Initialize session state
if 'paused' not in st.session_state:
    st.session_state.paused = False
    st.session_state.frame_rate = 0.2  # Adjusted default to reduce flicker
    st.session_state.frame_count = 0
    st.session_state.logs = []
    st.session_state.anomaly_counts = []
    st.session_state.frame_numbers = []
    st.session_state.total_detected = 0
    st.session_state.snapshots = []
    st.session_state.last_frame = None
    st.session_state.last_metrics = {}
    st.session_state.last_timestamp = ""

# Create snapshots directory
SNAPSHOT_FOLDER = "./snapshots"
os.makedirs(SNAPSHOT_FOLDER, exist_ok=True)

# Function to resize and pad image to 640x640 while preserving aspect ratio
def preprocess_image(image, target_size=(640, 640)):
    h, w = image.shape[:2]
    target_h, target_w = target_size

    # Calculate scaling factor to maintain aspect ratio
    scale = min(target_w / w, target_h / h)
    new_w, new_h = int(w * scale), int(h * scale)

    # Resize image
    resized_image = cv2.resize(image, (new_w, new_h), interpolation=cv2.INTER_AREA)

    # Create a blank 640x640 image
    padded_image = np.zeros((target_h, target_w, 3), dtype=np.uint8)

    # Calculate padding offsets
    top = (target_h - new_h) // 2
    left = (target_w - new_w) // 2

    # Place resized image in the center
    padded_image[top:top + new_h, left:left + new_w] = resized_image

    return padded_image

# Core monitor function
def monitor_feed(video_path, detection_type, model):
    if st.session_state.paused and st.session_state.last_frame is not None:
        frame = st.session_state.last_frame.copy()
        metrics = st.session_state.last_metrics.copy()
        num_anomalies = len(metrics.get('anomalies', []))
        snapshot_paths = [snapshot["path"] for snapshot in st.session_state.snapshots]
    else:
        cap = cv2.VideoCapture(video_path)
        cap.set(cv2.CAP_PROP_POS_FRAMES, st.session_state.frame_count)
        ret, frame = cap.read()
        cap.release()
        if not ret:
            return None, None, None, None, None

        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
        # Preprocess frame to 640x640
        frame_rgb = preprocess_image(frame_rgb, target_size=(640, 640))
        faults = detect_faults_solar(model, frame_rgb) if detection_type == "Solar Panel" else detect_faults_windmill(model, frame_rgb)
        num_anomalies = len(faults)

        # Draw bounding boxes and labels
        annotated_frame = frame_rgb.copy()
        for fault in faults:
            x, y = int(fault['location'][0]), int(fault['location'][1])
            cv2.rectangle(annotated_frame, (x-30, y-30), (x+30, y+30), (255, 0, 0), 2)
            cv2.putText(annotated_frame, f"{fault['type']}", (x, y-40), 
                        cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 0), 2)

        # Save snapshot if faults detected
        snapshot_paths = [snapshot["path"] for snapshot in st.session_state.snapshots]
        if faults:
            timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
            fault_types = "_".join([fault['type'].replace(" ", "_") for fault in faults])
            snapshot_filename = f"snapshot_{timestamp}_frame_{st.session_state.frame_count}_{fault_types}.png"
            snapshot_path = os.path.join(SNAPSHOT_FOLDER, snapshot_filename)
            cv2.imwrite(snapshot_path, cv2.cvtColor(annotated_frame, cv2.COLOR_RGB2BGR))
            st.session_state.snapshots.append({
                "path": snapshot_path,
                "log": f"{timestamp} - Frame {st.session_state.frame_count} - Anomalies: {num_anomalies} ({fault_types})"
            })
            if len(st.session_state.snapshots) > 5:
                st.session_state.snapshots.pop(0)
            snapshot_paths = [snapshot["path"] for snapshot in st.session_state.snapshots]

        st.session_state.frame_count += 1
        st.session_state.last_timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        st.session_state.last_frame = annotated_frame.copy()

        # Update metrics
        metrics = {"anomalies": faults, "total_detected": st.session_state.total_detected + num_anomalies}
        st.session_state.last_metrics = metrics.copy()
        st.session_state.total_detected += num_anomalies

        # Update logs and anomaly counts
        log_entry = f"{st.session_state.last_timestamp} - Frame {st.session_state.frame_count} - Anomalies: {num_anomalies}"
        st.session_state.logs.append(log_entry)
        st.session_state.anomaly_counts.append(num_anomalies)
        st.session_state.frame_numbers.append(st.session_state.frame_count)

        if len(st.session_state.logs) > 100:
            st.session_state.logs.pop(0)
            st.session_state.anomaly_counts.pop(0)
            st.session_state.frame_numbers.pop(0)

        frame = annotated_frame

    # Add frame count and timestamp to frame
    frame = cv2.resize(frame, (640, 480))
    cv2.putText(frame, f"Frame: {st.session_state.frame_count}", (10, 25), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
    cv2.putText(frame, f"{st.session_state.last_timestamp}", (10, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)

    metrics_str = "\n".join([f"{k}: {v}" for k, v in st.session_state.last_metrics.items()])
    logs_str = "\n".join(st.session_state.logs[-10:])

    # Generate chart
    fig, ax = plt.subplots(figsize=(6, 3))
    ax.plot(st.session_state.frame_numbers[-50:], st.session_state.anomaly_counts[-50:], marker='o', color='blue')
    ax.set_xlabel("Frame", fontsize=10)
    ax.set_ylabel("Count", fontsize=10)
    ax.grid(True)
    ax.tick_params(axis='both', which='major', labelsize=8)
    chart_path = "chart_temp.png"
    fig.savefig(chart_path)
    plt.close(fig)

    return frame, metrics_str, logs_str, chart_path, snapshot_paths

def main():
    st.markdown('<div class="main-header">THERMAL ANOMALY MONITORING DASHBOARD</div>', unsafe_allow_html=True)

    # Status banner
    status_placeholder = st.markdown(
        f'<div class="status">{"🟢 Running" if not st.session_state.paused else "⏸️ Paused"}</div>',
        unsafe_allow_html=True
    )

    # Sidebar for video selection and detection type
    st.sidebar.header("Settings")
    video_files = [f for f in os.listdir(VIDEO_FOLDER) if f.endswith('.mp4')]
    if not video_files:
        st.error("No videos found in the 'data' folder. Please add .mp4 files.")
        return
    video_file = st.sidebar.selectbox("Select Video", video_files)
    detection_type = st.sidebar.selectbox("Detection Type", ["Solar Panel", "Windmill"])

    # Load the appropriate model
    model = load_solar_model() if detection_type == "Solar Panel" else load_windmill_model()

    # Layout: Video feed and metrics
    col1, col2 = st.columns([3, 1])
    with col1:
        st.markdown('<div class="section-title">LIVE VIDEO FEED</div>', unsafe_allow_html=True)
        with st.container():
            st.markdown('<div class="section-box">', unsafe_allow_html=True)
            video_placeholder = st.empty()
            st.markdown('</div>', unsafe_allow_html=True)

    with col2:
        st.markdown('<div class="section-title">LIVE METRICS</div>', unsafe_allow_html=True)
        with st.container():
            st.markdown('<div class="section-box">', unsafe_allow_html=True)
            metrics_placeholder = st.empty()
            prediction_placeholder = st.empty()
            st.markdown('</div>', unsafe_allow_html=True)

    # Layout: Logs and trends
    col3, col4 = st.columns([1, 2])
    with col3:
        st.markdown('<div class="section-title">LIVE LOGS</div>', unsafe_allow_html=True)
        with st.container():
            st.markdown('<div class="section-box">', unsafe_allow_html=True)
            logs_placeholder = st.empty()
            st.markdown('</div>', unsafe_allow_html=True)

        st.markdown('<div class="section-title">LAST 5 CAPTURED EVENTS</div>', unsafe_allow_html=True)
        with st.container():
            st.markdown('<div class="section-box">', unsafe_allow_html=True)
            gallery_placeholder = st.empty()
            st.markdown('</div>', unsafe_allow_html=True)

    with col4:
        st.markdown('<div class="section-title">DETECTION TRENDS</div>', unsafe_allow_html=True)
        with st.container():
            st.markdown('<div class="section-box">', unsafe_allow_html=True)
            st.markdown('<div style="font-size: 14px; font-weight: bold; margin-bottom: 10px;">Anomalies Over Time</div>', unsafe_allow_html=True)
            trends_placeholder = st.empty()
            st.markdown('</div>', unsafe_allow_html=True)

    # Controls
    with st.container():
        col5, col6, col7 = st.columns([1, 1, 2])
        with col5:
            pause_btn = st.button("⏸️ Pause")
        with col6:
            resume_btn = st.button("▶️ Resume")
        with col7:
            frame_rate = st.slider("Frame Interval (seconds)", 0.1, 1.0, st.session_state.frame_rate)  # Adjusted range

    # Handle button clicks and slider
    if pause_btn:
        st.session_state.paused = True
        status_placeholder.markdown('<div class="status">⏸️ Paused</div>', unsafe_allow_html=True)
    if resume_btn:
        st.session_state.paused = False
        status_placeholder.markdown('<div class="status">🟢 Running</div>', unsafe_allow_html=True)
    st.session_state.frame_rate = frame_rate

    # Streaming loop
    video_path = os.path.join(VIDEO_FOLDER, video_file)
    while True:
        if st.session_state.paused:
            # Skip processing if paused and last frame exists
            if st.session_state.last_frame is not None:
                frame, metrics, logs, chart, snapshots = monitor_feed(video_path, detection_type, model)
                if frame is None:
                    st.success("Video processing completed.")
                    break
                # Update UI only if necessary
                video_placeholder.image(frame, channels="RGB", width=640)
                metrics_formatted = metrics.replace("\n", "<br>")
                metrics_placeholder.markdown(
                    f'<div class="metrics-text">{metrics_formatted}</div>',
                    unsafe_allow_html=True
                )
                logs_formatted = logs.replace("\n", "</div><div class='log-entry'>")
                logs_placeholder.markdown(
                    f'<div class="log-entry">{logs_formatted}</div>',
                    unsafe_allow_html=True
                )
                trends_placeholder.image(chart)
                with gallery_placeholder.container():
                    cols = st.columns(5)
                    for i, col in enumerate(cols):
                        with col:
                            if i < len(snapshots):
                                st.image(snapshots[i], width=100)
                                st.markdown(f'<div class="log-entry">{st.session_state.snapshots[i]["log"]}</div>', unsafe_allow_html=True)
                            else:
                                st.empty()
                prediction = predict_anomaly(st.session_state.anomaly_counts)
                if prediction:
                    prediction_placeholder.warning("**Prediction:** Potential issue detected - anomaly spike detected!")
                else:
                    prediction_placeholder.empty()
                time.sleep(st.session_state.frame_rate)
                continue  # Avoid rerun when paused
            else:
                frame, metrics, logs, chart, snapshots = monitor_feed(video_path, detection_type, model)
                if frame is None:
                    st.success("Video processing completed.")
                    break
        else:
            frame, metrics, logs, chart, snapshots = monitor_feed(video_path, detection_type, model)
            if frame is None:
                st.success("Video processing completed.")
                break

        # Update UI
        video_placeholder.image(frame, channels="RGB", width=640)
        metrics_formatted = metrics.replace("\n", "<br>")
        metrics_placeholder.markdown(
            f'<div class="metrics-text">{metrics_formatted}</div>',
            unsafe_allow_html=True
        )
        logs_formatted = logs.replace("\n", "</div><div class='log-entry'>")
        logs_placeholder.markdown(
            f'<div class="log-entry">{logs_formatted}</div>',
            unsafe_allow_html=True
        )
        trends_placeholder.image(chart)

        # Update gallery
        with gallery_placeholder.container():
            cols = st.columns(5)
            for i, col in enumerate(cols):
                with col:
                    if i < len(snapshots):
                        st.image(snapshots[i], width=100)
                        st.markdown(f'<div class="log-entry">{st.session_state.snapshots[i]["log"]}</div>', unsafe_allow_html=True)
                    else:
                        st.empty()

        # Predictive anomaly detection
        prediction = predict_anomaly(st.session_state.anomaly_counts)
        if prediction:
            prediction_placeholder.warning("**Prediction:** Potential issue detected - anomaly spike detected!")
        else:
            prediction_placeholder.empty()

        time.sleep(st.session_state.frame_rate)
        st.rerun()

if __name__ == "__main__":
    main()