app.py

import os
import cv2
import numpy as np
import torch
import torch.nn as nn
import torch.optim as optim
from facenet_pytorch import InceptionResnetV1, MTCNN
import mediapipe as mp
from fer import FER
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler, MinMaxScaler
from sklearn.metrics import silhouette_score
from scipy.spatial.distance import cdist
import umap
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.ticker import MaxNLocator
import gradio as gr
import tempfile

# Initialize models and other global variables
device = 'cuda' if torch.cuda.is_available() else 'cpu'

mtcnn = MTCNN(keep_all=False, device=device, thresholds=[0.999, 0.999, 0.999], min_face_size=100, selection_method='largest')
model = InceptionResnetV1(pretrained='vggface2').eval().to(device)
mp_face_mesh = mp.solutions.face_mesh
face_mesh = mp_face_mesh.FaceMesh(static_image_mode=False, max_num_faces=1, min_detection_confidence=0.5)
emotion_detector = FER(mtcnn=False)

def frame_to_timecode(frame_num, original_fps, desired_fps):
    total_seconds = frame_num / original_fps
    hours = int(total_seconds // 3600)
    minutes = int((total_seconds % 3600) // 60)
    seconds = int(total_seconds % 60)
    milliseconds = int((total_seconds - int(total_seconds)) * 1000)
    return f"{hours:02d}:{minutes:02d}:{seconds:02d}.{milliseconds:03d}"

def get_face_embedding_and_emotion(face_img):
    face_tensor = torch.tensor(face_img).permute(2, 0, 1).unsqueeze(0).float() / 255
    face_tensor = (face_tensor - 0.5) / 0.5
    face_tensor = face_tensor.to(device)
    with torch.no_grad():
        embedding = model(face_tensor)

    emotions = emotion_detector.detect_emotions(face_img)
    if emotions:
        emotion_dict = emotions[0]['emotions']
    else:
        emotion_dict = {e: 0 for e in ['angry', 'disgust', 'fear', 'happy', 'sad', 'surprise', 'neutral']}

    return embedding.cpu().numpy().flatten(), emotion_dict

def alignFace(img):
    img_raw = img.copy()
    results = face_mesh.process(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    if not results.multi_face_landmarks:
        return None
    landmarks = results.multi_face_landmarks[0].landmark
    left_eye = np.array([[landmarks[33].x, landmarks[33].y], [landmarks[160].x, landmarks[160].y],
                         [landmarks[158].x, landmarks[158].y], [landmarks[144].x, landmarks[144].y],
                         [landmarks[153].x, landmarks[153].y], [landmarks[145].x, landmarks[145].y]])
    right_eye = np.array([[landmarks[362].x, landmarks[362].y], [landmarks[385].x, landmarks[385].y],
                          [landmarks[387].x, landmarks[387].y], [landmarks[263].x, landmarks[263].y],
                          [landmarks[373].x, landmarks[373].y], [landmarks[380].x, landmarks[380].y]])
    left_eye_center = left_eye.mean(axis=0).astype(np.int32)
    right_eye_center = right_eye.mean(axis=0).astype(np.int32)
    dY = right_eye_center[1] - left_eye_center[1]
    dX = right_eye_center[0] - left_eye_center[0]
    angle = np.degrees(np.arctan2(dY, dX))
    desired_angle = 0
    angle_diff = desired_angle - angle
    height, width = img_raw.shape[:2]
    center = (width // 2, height // 2)
    rotation_matrix = cv2.getRotationMatrix2D(center, angle_diff, 1)
    new_img = cv2.warpAffine(img_raw, rotation_matrix, (width, height))
    return new_img

def extract_and_align_faces_from_video(video_path, aligned_faces_folder, desired_fps):
    video = cv2.VideoCapture(video_path)
    if not video.isOpened():
        print(f"Error: Could not open video file at {video_path}")
        return {}, {}, desired_fps, 0
    frame_count = int(video.get(cv2.CAP_PROP_FRAME_COUNT))
    original_fps = video.get(cv2.CAP_PROP_FPS)
    if frame_count == 0:
        print(f"Error: Video file at {video_path} appears to be empty")
        return {}, {}, desired_fps, 0
    embeddings_by_frame = {}
    emotions_by_frame = {}

    for frame_num in range(0, frame_count, int(original_fps / desired_fps)):
        video.set(cv2.CAP_PROP_POS_FRAMES, frame_num)
        ret, frame = video.read()
        if not ret or frame is None:
            print(f"Error: Could not read frame {frame_num}")
            continue
        try:
            boxes, probs = mtcnn.detect(frame)
            if boxes is not None and len(boxes) > 0:
                box = boxes[0]
                if probs[0] >= 0.99:
                    x1, y1, x2, y2 = [int(b) for b in box]
                    face = frame[y1:y2, x1:x2]
                    aligned_face = alignFace(face)
                    if aligned_face is not None:
                        aligned_face_resized = cv2.resize(aligned_face, (160, 160))
                        output_path = os.path.join(aligned_faces_folder, f"frame_{frame_num}_face.jpg")
                        cv2.imwrite(output_path, aligned_face_resized)
                        embedding, emotion = get_face_embedding_and_emotion(aligned_face_resized)
                        embeddings_by_frame[frame_num] = embedding
                        emotions_by_frame[frame_num] = emotion
        except Exception as e:
            print(f"Error processing frame {frame_num}: {str(e)}")
            continue

    video.release()
    return embeddings_by_frame, emotions_by_frame, desired_fps, original_fps

def cluster_embeddings(embeddings):
    if len(embeddings) < 2:
        print("Not enough embeddings for clustering. Assigning all to one cluster.")
        return np.zeros(len(embeddings), dtype=int)
    n_clusters = min(3, len(embeddings))  # Use at most 3 clusters
    scaler = StandardScaler()
    embeddings_scaled = scaler.fit_transform(embeddings)
    kmeans = KMeans(n_clusters=n_clusters, random_state=42, n_init=10)
    clusters = kmeans.fit_predict(embeddings_scaled)
    return clusters

def organize_faces_by_person(embeddings_by_frame, clusters, aligned_faces_folder, organized_faces_folder):
    for (frame_num, embedding), cluster in zip(embeddings_by_frame.items(), clusters):
        person_folder = os.path.join(organized_faces_folder, f"person_{cluster}")
        os.makedirs(person_folder, exist_ok=True)
        src = os.path.join(aligned_faces_folder, f"frame_{frame_num}_face.jpg")
        dst = os.path.join(person_folder, f"frame_{frame_num}_face.jpg")
        shutil.copy(src, dst)

def save_person_data_to_csv(embeddings_by_frame, emotions_by_frame, clusters, desired_fps, original_fps, output_folder, num_components):
    emotions = ['angry', 'disgust', 'fear', 'happy', 'sad', 'neutral']
    person_data = {}

    for (frame_num, embedding), (_, emotion_dict), cluster in zip(embeddings_by_frame.items(),
                                                                  emotions_by_frame.items(), clusters):
        if cluster not in person_data:
            person_data[cluster] = []
        person_data[cluster].append((frame_num, embedding, {e: emotion_dict[e] for e in emotions}))

    largest_cluster = max(person_data, key=lambda k: len(person_data[k]))

    data = person_data[largest_cluster]
    data.sort(key=lambda x: x[0])
    frames, embeddings, emotions_data = zip(*data)

    embeddings_array = np.array(embeddings)
    np.save(os.path.join(output_folder, 'face_embeddings.npy'), embeddings_array)

    reducer = umap.UMAP(n_components=num_components, random_state=1)
    embeddings_reduced = reducer.fit_transform(embeddings)

    scaler = MinMaxScaler(feature_range=(0, 1))
    embeddings_reduced_normalized = scaler.fit_transform(embeddings_reduced)

    timecodes = [frame_to_timecode(frame, original_fps, desired_fps) for frame in frames]
    times_in_minutes = [frame / (original_fps * 60) for frame in frames]

    df_data = {
        'Frame': frames,
        'Timecode': timecodes,
        'Time (Minutes)': times_in_minutes,
        'Embedding_Index': range(len(embeddings))
    }

    for i in range(num_components):
        df_data[f'Comp {i + 1}'] = embeddings_reduced_normalized[:, i]

    for emotion in emotions:
        df_data[emotion] = [e[emotion] for e in emotions_data]

    df = pd.DataFrame(df_data)

    return df, largest_cluster

class LSTMAutoencoder(nn.Module):
    def __init__(self, input_size, hidden_size=64, num_layers=2):
        super(LSTMAutoencoder, self).__init__()
        self.input_size = input_size
        self.hidden_size = hidden_size
        self.num_layers = num_layers

        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_size, input_size)

    def forward(self, x):
        _, (hidden, _) = self.lstm(x)
        out = self.fc(hidden[-1])
        return out

def lstm_anomaly_detection(X, feature_columns, num_anomalies=10, epochs=100, batch_size=64):
    device = 'cuda' if torch.cuda.is_available() else 'cpu'

    X = torch.FloatTensor(X).to(device)

    train_size = int(0.85 * len(X))
    X_train, X_val = X[:train_size], X[train_size:]

    model = LSTMAutoencoder(input_size=len(feature_columns)).to(device)
    criterion = nn.MSELoss()
    optimizer = optim.Adam(model.parameters())

    for epoch in range(epochs):
        model.train()
        optimizer.zero_grad()
        output_train = model(X_train.unsqueeze(0))
        loss_train = criterion(output_train, X_train)
        loss_train.backward()
        optimizer.step()

        model.eval()
        with torch.no_grad():
            output_val = model(X_val.unsqueeze(0))
            loss_val = criterion(output_val, X_val)

    model.eval()
    with torch.no_grad():
        reconstructed = model(X.unsqueeze(0)).squeeze(0).cpu().numpy()

    mse = np.mean(np.power(X.cpu().numpy() - reconstructed, 2), axis=1)

    top_indices = mse.argsort()[-num_anomalies:][::-1]
    anomalies = np.zeros(len(mse), dtype=bool)
    anomalies[top_indices] = True

    return anomalies, mse, top_indices, model

def plot_anomaly_scores(df, anomaly_scores, top_indices, title):
    fig, ax = plt.subplots(figsize=(16, 8))
    bars = ax.bar(range(len(df)), anomaly_scores, width=0.8)
    for i in top_indices:
        bars[i].set_color('red')
    ax.set_xlabel('Timecode')
    ax.set_ylabel('Anomaly Score')
    ax.set_title(f'Anomaly Scores Over Time ({title})')
    ax.xaxis.set_major_locator(MaxNLocator(nbins=100))
    ticks = ax.get_xticks()
    ax.set_xticklabels([df['Timecode'].iloc[int(tick)] if tick >= 0 and tick < len(df) else '' for tick in ticks], rotation=90, ha='right')
    plt.tight_layout()
    return fig

def plot_emotion(df, emotion):
    fig, ax = plt.subplots(figsize=(16, 8))
    values = df[emotion].values
    bars = ax.bar(range(len(df)), values, width=0.8)
    top_10_indices = np.argsort(values)[-10:]
    for i, bar in enumerate(bars):
        if i in top_10_indices:
            bar.set_color('red')
    ax.set_xlabel('Timecode')
    ax.set_ylabel(f'{emotion.capitalize()} Score')
    ax.set_title(f'{emotion.capitalize()} Scores Over Time')
    ax.xaxis.set_major_locator(MaxNLocator(nbins=100))
    ticks = ax.get_xticks()
    ax.set_xticklabels([df['Timecode'].iloc[int(tick)] if tick >= 0 and tick < len(df) else '' for tick in ticks], rotation=90, ha='right')
    plt.tight_layout()
    return fig

def process_video(video_path, num_anomalies, num_components, desired_fps, batch_size, progress=gr.Progress()):
    with tempfile.TemporaryDirectory() as temp_dir:
        aligned_faces_folder = os.path.join(temp_dir, 'aligned_faces')
        organized_faces_folder = os.path.join(temp_dir, 'organized_faces')
        os.makedirs(aligned_faces_folder, exist_ok=True)
        os.makedirs(organized_faces_folder, exist_ok=True)

        progress(0.1, "Extracting and aligning faces")
        embeddings_by_frame, emotions_by_frame, _, original_fps = extract_and_align_faces_from_video(video_path, aligned_faces_folder, desired_fps)

        if not embeddings_by_frame:
            return "No faces were extracted from the video.", None, None, None, None

        progress(0.3, "Clustering embeddings")
        embeddings = list(embeddings_by_frame.values())
        clusters = cluster_embeddings(embeddings)

        progress(0.4, "Organizing faces")
        organize_faces_by_person(embeddings_by_frame, clusters, aligned_faces_folder, organized_faces_folder)

        progress(0.5, "Saving person data")
        df, largest_cluster = save_person_data_to_csv(embeddings_by_frame, emotions_by_frame, clusters, desired_fps, original_fps, temp_dir, num_components)

        progress(0.6, "Performing anomaly detection")
        feature_columns = [col for col in df.columns if col not in ['Frame', 'Timecode', 'Time (Minutes)', 'Embedding_Index']]
        anomalies_all, anomaly_scores_all, top_indices_all, _ = lstm_anomaly_detection(df[feature_columns].values, feature_columns, num_anomalies=num_anomalies, batch_size=batch_size)
        
        progress(0.8, "Generating plots")
        anomaly_plot = plot_anomaly_scores(df, anomaly_scores_all, top_indices_all, "All Features")
        emotion_plots = [plot_emotion(df, emotion) for emotion in ['fear', 'sad', 'angry']]

        progress(0.9, "Preparing results")
        results = f"Top {num_anomalies} anomalies (All Features):\n"
        results += "\n".join([f"{score:.4f} at {timecode}" for score, timecode in 
                              zip(anomaly_scores_all[top_indices_all], df['Timecode'].iloc[top_indices_all].values)])

        progress(1.0, "Complete")
        return results, anomaly_plot, *emotion_plots

# Gradio interface
iface = gr.Interface(
    fn=process_video,
    inputs=[
        gr.Video(),
        gr.Slider(minimum=1, maximum=20, step=1, value=10, label="Number of Anomalies"),
        gr.Slider(minimum=2, maximum=5, step=1, value=3, label="Number of Components"),
        gr.Slider(minimum=1, maximum=30, step=1, value=20, label="Desired FPS"),
        gr.Slider(minimum=1, maximum=64, step=1, value=16, label="Batch Size")
    ],
    outputs=[
        gr.Textbox(label="Anomaly Detection Results"),
        gr.Plot(label="Anomaly Scores"),
        gr.Plot(label="Fear Scores"),
        gr.Plot(label="Sad Scores"),
        gr.Plot(label="Angry Scores")
    ],
    title="Video Anomaly Detection",
    description="Upload a video to detect anomalies in facial expressions and emotions. Adjust parameters as needed."
)

iface.launch()