|
import math |
|
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 tensorflow as tf |
|
import mediapipe as mp |
|
from fer import FER |
|
from sklearn.cluster import DBSCAN |
|
from sklearn.preprocessing import StandardScaler, MinMaxScaler |
|
import pandas as pd |
|
import matplotlib |
|
import matplotlib.pyplot as plt |
|
from matplotlib.patches import Rectangle |
|
from moviepy.editor import VideoFileClip |
|
from PIL import Image |
|
import gradio as gr |
|
import tempfile |
|
import shutil |
|
import copy |
|
import time |
|
|
|
matplotlib.rcParams['figure.dpi'] = 500 |
|
matplotlib.rcParams['savefig.dpi'] = 500 |
|
|
|
|
|
device = 'cuda' if torch.cuda.is_available() else 'cpu' |
|
|
|
mtcnn = MTCNN(keep_all=False, device=device, thresholds=[0.95, 0.95, 0.95], min_face_size=80) |
|
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, total_frames, duration): |
|
total_seconds = (frame_num / total_frames) * duration |
|
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 seconds_to_timecode(seconds): |
|
hours = int(seconds // 3600) |
|
minutes = int((seconds % 3600) // 60) |
|
seconds = int(seconds % 60) |
|
return f"{hours:02d}:{minutes:02d}:{seconds:02d}" |
|
|
|
def timecode_to_seconds(timecode): |
|
h, m, s = map(int, timecode.split(':')) |
|
return h * 3600 + m * 60 + s |
|
|
|
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', 'sad', 'happy']} |
|
|
|
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_frames(video_path, output_folder, desired_fps, progress_callback=None): |
|
os.makedirs(output_folder, exist_ok=True) |
|
clip = VideoFileClip(video_path) |
|
original_fps = clip.fps |
|
duration = clip.duration |
|
total_frames = int(duration * original_fps) |
|
step = max(1, original_fps / desired_fps) |
|
total_frames_to_extract = int(total_frames / step) |
|
|
|
frame_count = 0 |
|
for t in np.arange(0, duration, step / original_fps): |
|
frame = clip.get_frame(t) |
|
img = Image.fromarray(frame) |
|
img.save(os.path.join(output_folder, f"frame_{frame_count:04d}.jpg")) |
|
frame_count += 1 |
|
if progress_callback: |
|
progress = min(100, (frame_count / total_frames_to_extract) * 100) |
|
progress_callback(progress, f"Extracting frame") |
|
if frame_count >= total_frames_to_extract: |
|
break |
|
clip.close() |
|
return frame_count, original_fps |
|
|
|
def is_frontal_face(landmarks, threshold=40): |
|
nose_tip = landmarks[4] |
|
left_chin = landmarks[234] |
|
right_chin = landmarks[454] |
|
nose_to_left = [left_chin.x - nose_tip.x, left_chin.y - nose_tip.y] |
|
nose_to_right = [right_chin.x - nose_tip.x, right_chin.y - nose_tip.y] |
|
dot_product = nose_to_left[0] * nose_to_right[0] + nose_to_left[1] * nose_to_right[1] |
|
magnitude_left = math.sqrt(nose_to_left[0] ** 2 + nose_to_left[1] ** 2) |
|
magnitude_right = math.sqrt(nose_to_right[0] ** 2 + nose_to_right[1] ** 2) |
|
cos_angle = dot_product / (magnitude_left * magnitude_right) |
|
angle = math.acos(cos_angle) |
|
angle_degrees = math.degrees(angle) |
|
return abs(180 - angle_degrees) < threshold |
|
|
|
def process_frames(frames_folder, aligned_faces_folder, frame_count, progress, batch_size): |
|
embeddings_by_frame = {} |
|
emotions_by_frame = {} |
|
aligned_face_paths = [] |
|
frame_files = sorted([f for f in os.listdir(frames_folder) if f.endswith('.jpg')]) |
|
|
|
for i in range(0, len(frame_files), batch_size): |
|
batch_files = frame_files[i:i + batch_size] |
|
batch_frames = [] |
|
batch_nums = [] |
|
|
|
for frame_file in batch_files: |
|
frame_num = int(frame_file.split('_')[1].split('.')[0]) |
|
frame_path = os.path.join(frames_folder, frame_file) |
|
frame = cv2.imread(frame_path) |
|
if frame is not None: |
|
batch_frames.append(frame) |
|
batch_nums.append(frame_num) |
|
|
|
if batch_frames: |
|
batch_boxes, batch_probs = mtcnn.detect(batch_frames) |
|
|
|
for j, (frame, frame_num, boxes, probs) in enumerate( |
|
zip(batch_frames, batch_nums, batch_boxes, batch_probs)): |
|
if boxes is not None and len(boxes) > 0 and probs[0] >= 0.99: |
|
x1, y1, x2, y2 = [int(b) for b in boxes[0]] |
|
face = frame[y1:y2, x1:x2] |
|
if face.size > 0: |
|
results = face_mesh.process(cv2.cvtColor(face, cv2.COLOR_BGR2RGB)) |
|
if results.multi_face_landmarks and is_frontal_face(results.multi_face_landmarks[0].landmark): |
|
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) |
|
aligned_face_paths.append(output_path) |
|
embedding, emotion = get_face_embedding_and_emotion(aligned_face_resized) |
|
embeddings_by_frame[frame_num] = embedding |
|
emotions_by_frame[frame_num] = emotion |
|
|
|
progress((i + len(batch_files)) / len(frame_files), |
|
f"Processing frames {i + 1} to {min(i + len(batch_files), len(frame_files))} of {len(frame_files)}") |
|
|
|
return embeddings_by_frame, emotions_by_frame, aligned_face_paths |
|
|
|
def cluster_faces(embeddings): |
|
if len(embeddings) < 2: |
|
print("Not enough faces for clustering. Assigning all to one cluster.") |
|
return np.zeros(len(embeddings), dtype=int) |
|
|
|
X = np.stack(embeddings) |
|
dbscan = DBSCAN(eps=0.5, min_samples=5, metric='cosine') |
|
clusters = dbscan.fit_predict(X) |
|
|
|
if np.all(clusters == -1): |
|
print("DBSCAN assigned all to noise. Considering as one cluster.") |
|
return np.zeros(len(embeddings), dtype=int) |
|
|
|
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, video_duration): |
|
emotions = ['angry', 'disgust', 'fear', 'sad', 'happy'] |
|
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) |
|
|
|
total_frames = max(frames) |
|
timecodes = [frame_to_timecode(frame, total_frames, video_duration) for frame in frames] |
|
|
|
df_data = { |
|
'Frame': frames, |
|
'Timecode': timecodes, |
|
'Embedding_Index': range(len(embeddings)) |
|
} |
|
|
|
for i in range(len(embeddings[0])): |
|
df_data[f'Raw_Embedding_{i}'] = [embedding[i] for embedding in embeddings] |
|
|
|
for emotion in emotions: |
|
df_data[emotion] = [e[emotion] for e in emotions_data] |
|
|
|
df = pd.DataFrame(df_data) |
|
|
|
return df, largest_cluster |
|
|
|
class Autoencoder(nn.Module): |
|
def __init__(self, input_size): |
|
super(Autoencoder, self).__init__() |
|
self.encoder = nn.Sequential( |
|
nn.Linear(input_size, 512), |
|
nn.ReLU(), |
|
nn.Linear(512, 256), |
|
nn.ReLU(), |
|
nn.Linear(256, 128), |
|
nn.ReLU(), |
|
nn.Linear(128, 64) |
|
) |
|
self.decoder = nn.Sequential( |
|
nn.Linear(64, 128), |
|
nn.ReLU(), |
|
nn.Linear(128, 256), |
|
nn.ReLU(), |
|
nn.Linear(256, 512), |
|
nn.ReLU(), |
|
nn.Linear(512, input_size) |
|
) |
|
|
|
def forward(self, x): |
|
batch_size, seq_len, _ = x.size() |
|
x = x.view(batch_size * seq_len, -1) |
|
encoded = self.encoder(x) |
|
decoded = self.decoder(encoded) |
|
return decoded.view(batch_size, seq_len, -1) |
|
|
|
def determine_anomalies(mse_values, threshold): |
|
mean = np.mean(mse_values) |
|
std = np.std(mse_values) |
|
anomalies = mse_values > (mean + threshold * std) |
|
return anomalies |
|
|
|
def anomaly_detection(X_emotions, X_embeddings, epochs=200, batch_size=8, patience=3): |
|
device = 'cuda' if torch.cuda.is_available() else 'cpu' |
|
|
|
|
|
scaler_emotions = MinMaxScaler() |
|
X_emotions_scaled = scaler_emotions.fit_transform(X_emotions) |
|
|
|
|
|
X_emotions_scaled = torch.FloatTensor(X_emotions_scaled).to(device) |
|
if X_emotions_scaled.dim() == 2: |
|
X_emotions_scaled = X_emotions_scaled.unsqueeze(0) |
|
|
|
model_emotions = Autoencoder(input_size=X_emotions_scaled.shape[2]).to(device) |
|
criterion = nn.MSELoss() |
|
optimizer_emotions = optim.Adam(model_emotions.parameters()) |
|
|
|
|
|
for epoch in range(epochs): |
|
model_emotions.train() |
|
optimizer_emotions.zero_grad() |
|
output_emotions = model_emotions(X_emotions_scaled) |
|
loss_emotions = criterion(output_emotions, X_emotions_scaled) |
|
loss_emotions.backward() |
|
optimizer_emotions.step() |
|
|
|
|
|
X_embeddings = torch.FloatTensor(X_embeddings).to(device) |
|
if X_embeddings.dim() == 2: |
|
X_embeddings = X_embeddings.unsqueeze(0) |
|
|
|
model_embeddings = Autoencoder(input_size=X_embeddings.shape[2]).to(device) |
|
optimizer_embeddings = optim.Adam(model_embeddings.parameters()) |
|
|
|
|
|
for epoch in range(epochs): |
|
model_embeddings.train() |
|
optimizer_embeddings.zero_grad() |
|
output_embeddings = model_embeddings(X_embeddings) |
|
loss_embeddings = criterion(output_embeddings, X_embeddings) |
|
loss_embeddings.backward() |
|
optimizer_embeddings.step() |
|
|
|
|
|
model_emotions.eval() |
|
model_embeddings.eval() |
|
with torch.no_grad(): |
|
reconstructed_emotions = model_emotions(X_emotions_scaled).cpu().numpy() |
|
reconstructed_embeddings = model_embeddings(X_embeddings).cpu().numpy() |
|
|
|
mse_emotions = np.mean(np.power(X_emotions_scaled.cpu().numpy() - reconstructed_emotions, 2), axis=2).squeeze() |
|
mse_embeddings = np.mean(np.power(X_embeddings.cpu().numpy() - reconstructed_embeddings, 2), axis=2).squeeze() |
|
|
|
return mse_emotions, mse_embeddings |
|
|
|
def plot_mse(df, mse_values, title, color='blue', time_threshold=3, anomaly_threshold=4): |
|
plt.figure(figsize=(16, 8), dpi=500) |
|
fig, ax = plt.subplots(figsize=(16, 8)) |
|
|
|
if 'Seconds' not in df.columns: |
|
df['Seconds'] = df['Timecode'].apply( |
|
lambda x: sum(float(t) * 60 ** i for i, t in enumerate(reversed(x.split(':'))))) |
|
|
|
|
|
min_length = min(len(df), len(mse_values)) |
|
df = df.iloc[:min_length] |
|
mse_values = mse_values[:min_length] |
|
|
|
|
|
mask = ~np.isnan(mse_values) |
|
df = df[mask] |
|
mse_values = mse_values[mask] |
|
|
|
mean = pd.Series(mse_values).rolling(window=10).mean() |
|
std = pd.Series(mse_values).rolling(window=10).std() |
|
median = np.median(mse_values) |
|
|
|
ax.scatter(df['Seconds'], mse_values, color=color, alpha=0.3, s=5) |
|
ax.plot(df['Seconds'], mean, color=color, linewidth=2) |
|
ax.fill_between(df['Seconds'], mean - std, mean + std, color=color, alpha=0.2) |
|
|
|
|
|
ax.axhline(y=median, color='black', linestyle='--', label='Baseline') |
|
ax.text(ax.get_xlim()[1], median, 'Baseline', verticalalignment='center', horizontalalignment='left', color='black') |
|
|
|
|
|
threshold = np.mean(mse_values) + anomaly_threshold * np.std(mse_values) |
|
ax.axhline(y=threshold, color='red', linestyle='--', label=f'Threshold: {anomaly_threshold:.1f}') |
|
ax.text(ax.get_xlim()[1], threshold, f'Threshold: {anomaly_threshold:.1f}', verticalalignment='center', horizontalalignment='left', color='red') |
|
|
|
anomalies = determine_anomalies(mse_values, anomaly_threshold) |
|
anomaly_frames = df['Frame'].iloc[anomalies].tolist() |
|
|
|
ax.scatter(df['Seconds'].iloc[anomalies], mse_values[anomalies], color='red', s=25, zorder=5) |
|
|
|
anomaly_data = list(zip(df['Timecode'].iloc[anomalies], |
|
df['Seconds'].iloc[anomalies], |
|
mse_values[anomalies])) |
|
anomaly_data.sort(key=lambda x: x[1]) |
|
|
|
grouped_anomalies = [] |
|
current_group = [] |
|
for timecode, sec, mse in anomaly_data: |
|
if not current_group or sec - current_group[-1][1] <= time_threshold: |
|
current_group.append((timecode, sec, mse)) |
|
else: |
|
grouped_anomalies.append(current_group) |
|
current_group = [(timecode, sec, mse)] |
|
if current_group: |
|
grouped_anomalies.append(current_group) |
|
|
|
for group in grouped_anomalies: |
|
start_sec = group[0][1] |
|
end_sec = group[-1][1] |
|
rect = Rectangle((start_sec, ax.get_ylim()[0]), end_sec - start_sec, ax.get_ylim()[1] - ax.get_ylim()[0], |
|
facecolor='red', alpha=0.3, zorder=1) |
|
ax.add_patch(rect) |
|
|
|
for group in grouped_anomalies: |
|
highest_mse_anomaly = max(group, key=lambda x: x[2]) |
|
timecode, sec, mse = highest_mse_anomaly |
|
ax.annotate(timecode, (sec, mse), textcoords="offset points", xytext=(0, 10), |
|
ha='center', fontsize=6, color='red') |
|
|
|
max_seconds = df['Seconds'].max() |
|
num_ticks = 100 |
|
tick_locations = np.linspace(0, max_seconds, num_ticks) |
|
tick_labels = [seconds_to_timecode(int(s)) for s in tick_locations] |
|
|
|
ax.set_xticks(tick_locations) |
|
ax.set_xticklabels(tick_labels, rotation=90, ha='center', fontsize=6) |
|
|
|
ax.set_xlabel('Timecode') |
|
ax.set_ylabel('Mean Squared Error') |
|
ax.set_title(title) |
|
|
|
ax.grid(True, linestyle='--', alpha=0.7) |
|
ax.legend() |
|
plt.tight_layout() |
|
plt.close() |
|
return fig, anomaly_frames |
|
|
|
def plot_mse_histogram(mse_values, title, anomaly_threshold, color='blue'): |
|
plt.figure(figsize=(16, 8), dpi=500) |
|
fig, ax = plt.subplots(figsize=(16, 8)) |
|
|
|
ax.hist(mse_values, bins=100, edgecolor='black', color=color, alpha=0.7) |
|
ax.set_xlabel('Mean Squared Error') |
|
ax.set_ylabel('Number of Samples') |
|
ax.set_title(title) |
|
|
|
mean = np.mean(mse_values) |
|
std = np.std(mse_values) |
|
threshold = mean + anomaly_threshold * std |
|
|
|
ax.axvline(x=threshold, color='red', linestyle='--', linewidth=2) |
|
|
|
|
|
ax.annotate(f'Threshold: {anomaly_threshold:.1f}', |
|
xy=(threshold, ax.get_ylim()[0]), |
|
xytext=(0, -20), |
|
textcoords='offset points', |
|
ha='center', va='top', |
|
bbox=dict(boxstyle='round,pad=0.5', fc='white', ec='none', alpha=0.7), |
|
color='red') |
|
|
|
plt.tight_layout() |
|
plt.close() |
|
return fig |
|
|
|
|
|
def plot_emotion(df, emotion, color, anomaly_threshold): |
|
plt.figure(figsize=(16, 8), dpi=500) |
|
fig, ax = plt.subplots(figsize=(16, 8)) |
|
|
|
df['Seconds'] = df['Timecode'].apply( |
|
lambda x: sum(float(t) * 60 ** i for i, t in enumerate(reversed(x.split(':'))))) |
|
|
|
mean = df[emotion].rolling(window=10).mean() |
|
std = df[emotion].rolling(window=10).std() |
|
median = df[emotion].median() |
|
|
|
ax.scatter(df['Seconds'], df[emotion], color=color, alpha=0.3, s=5) |
|
ax.plot(df['Seconds'], mean, color=color, linewidth=2) |
|
ax.fill_between(df['Seconds'], mean - std, mean + std, color=color, alpha=0.2) |
|
|
|
|
|
ax.axhline(y=median, color='black', linestyle='--', label='Baseline') |
|
ax.text(ax.get_xlim()[1], median, 'Baseline', verticalalignment='center', horizontalalignment='left', color='black') |
|
|
|
|
|
probability_threshold = (anomaly_threshold - 1) / 6 |
|
|
|
|
|
ax.axhline(y=probability_threshold, color='red', linestyle='--', label=f'Threshold: {probability_threshold:.2f}') |
|
ax.text(ax.get_xlim()[1], probability_threshold, f'Threshold: {probability_threshold:.2f}', |
|
verticalalignment='center', horizontalalignment='left', color='red') |
|
|
|
|
|
anomalies = df[emotion] >= probability_threshold |
|
ax.scatter(df['Seconds'][anomalies], df[emotion][anomalies], color='red', s=25, zorder=5) |
|
|
|
max_seconds = df['Seconds'].max() |
|
num_ticks = 100 |
|
tick_locations = np.linspace(0, max_seconds, num_ticks) |
|
tick_labels = [seconds_to_timecode(int(s)) for s in tick_locations] |
|
|
|
ax.set_xticks(tick_locations) |
|
ax.set_xticklabels(tick_labels, rotation=90, ha='center', fontsize=6) |
|
|
|
ax.set_xlabel('Timecode') |
|
ax.set_ylabel('Emotion Probability') |
|
ax.set_title(f"{emotion.capitalize()} Over Time") |
|
|
|
ax.grid(True, linestyle='--', alpha=0.7) |
|
ax.legend() |
|
plt.tight_layout() |
|
plt.close() |
|
return fig |
|
|
|
def get_all_face_samples(organized_faces_folder, output_folder, largest_cluster, max_samples=500): |
|
face_samples = {"most_frequent": [], "others": []} |
|
for cluster_folder in sorted(os.listdir(organized_faces_folder)): |
|
if cluster_folder.startswith("person_"): |
|
person_folder = os.path.join(organized_faces_folder, cluster_folder) |
|
face_files = sorted([f for f in os.listdir(person_folder) if f.endswith('.jpg')]) |
|
if face_files: |
|
cluster_id = int(cluster_folder.split('_')[1]) |
|
if cluster_id == largest_cluster: |
|
for i, sample in enumerate(face_files[:max_samples]): |
|
face_path = os.path.join(person_folder, sample) |
|
output_path = os.path.join(output_folder, f"face_sample_most_frequent_{i:04d}.jpg") |
|
face_img = cv2.imread(face_path) |
|
if face_img is not None: |
|
small_face = cv2.resize(face_img, (160, 160)) |
|
cv2.imwrite(output_path, small_face) |
|
face_samples["most_frequent"].append(output_path) |
|
if len(face_samples["most_frequent"]) >= max_samples: |
|
break |
|
else: |
|
remaining_samples = max_samples - len(face_samples["others"]) |
|
if remaining_samples > 0: |
|
for i, sample in enumerate(face_files[:remaining_samples]): |
|
face_path = os.path.join(person_folder, sample) |
|
output_path = os.path.join(output_folder, f"face_sample_other_{cluster_id:02d}_{i:04d}.jpg") |
|
face_img = cv2.imread(face_path) |
|
if face_img is not None: |
|
small_face = cv2.resize(face_img, (160, 160)) |
|
cv2.imwrite(output_path, small_face) |
|
face_samples["others"].append(output_path) |
|
if len(face_samples["others"]) >= max_samples: |
|
break |
|
return face_samples |
|
|
|
def process_video(video_path, anomaly_threshold, desired_fps, progress=gr.Progress()): |
|
start_time = time.time() |
|
output_folder = "output" |
|
os.makedirs(output_folder, exist_ok=True) |
|
batch_size = 16 |
|
|
|
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) |
|
|
|
clip = VideoFileClip(video_path) |
|
video_duration = clip.duration |
|
clip.close() |
|
|
|
progress(0, "Starting frame extraction") |
|
frames_folder = os.path.join(temp_dir, 'extracted_frames') |
|
|
|
def extraction_progress(percent, message): |
|
progress(percent / 100, f"Extracting frames") |
|
|
|
frame_count, original_fps = extract_frames(video_path, frames_folder, desired_fps, extraction_progress) |
|
|
|
progress(1, "Frame extraction complete") |
|
progress(0.3, "Processing frames") |
|
embeddings_by_frame, emotions_by_frame, aligned_face_paths = process_frames(frames_folder, aligned_faces_folder, |
|
frame_count, |
|
progress, batch_size) |
|
|
|
if not aligned_face_paths: |
|
return ("No faces were extracted from the video.",) + (None,) * 10 |
|
|
|
progress(0.6, "Clustering faces") |
|
embeddings = [embedding for _, embedding in embeddings_by_frame.items()] |
|
clusters = cluster_faces(embeddings) |
|
num_clusters = len(set(clusters)) |
|
|
|
progress(0.7, "Organizing faces") |
|
organize_faces_by_person(embeddings_by_frame, clusters, aligned_faces_folder, organized_faces_folder) |
|
|
|
progress(0.8, "Saving person data") |
|
df, largest_cluster = save_person_data_to_csv(embeddings_by_frame, emotions_by_frame, clusters, desired_fps, |
|
original_fps, temp_dir, video_duration) |
|
|
|
|
|
df['Seconds'] = df['Timecode'].apply( |
|
lambda x: sum(float(t) * 60 ** i for i, t in enumerate(reversed(x.split(':'))))) |
|
|
|
progress(0.85, "Getting face samples") |
|
face_samples = get_all_face_samples(organized_faces_folder, output_folder, largest_cluster) |
|
|
|
progress(0.9, "Performing anomaly detection") |
|
emotion_columns = ['angry', 'disgust', 'fear', 'sad', 'happy'] |
|
embedding_columns = [col for col in df.columns if col.startswith('Raw_Embedding_')] |
|
|
|
X_emotions = df[emotion_columns].values |
|
X_embeddings = df[embedding_columns].values |
|
|
|
try: |
|
mse_emotions, mse_embeddings = anomaly_detection(X_emotions, X_embeddings, batch_size=batch_size) |
|
|
|
progress(0.95, "Generating plots") |
|
mse_plot_embeddings, anomaly_frames_embeddings = plot_mse(df, mse_embeddings, "Facial Embeddings", |
|
color='green', |
|
anomaly_threshold=anomaly_threshold) |
|
mse_histogram_embeddings = plot_mse_histogram(mse_embeddings, "MSE Distribution: Facial Embeddings", |
|
anomaly_threshold, color='green') |
|
|
|
|
|
emotion_plots = [] |
|
for emotion, color in zip(emotion_columns, ['purple', 'brown', 'green', 'orange', 'darkblue']): |
|
emotion_plot = plot_emotion(df, emotion, color, anomaly_threshold) |
|
emotion_plots.append(emotion_plot) |
|
|
|
mse_var_emotions = np.var(mse_emotions) |
|
mse_var_embeddings = np.var(mse_embeddings) |
|
|
|
except Exception as e: |
|
print(f"Error details: {str(e)}") |
|
return (f"Error in anomaly detection: {str(e)}",) + (None,) * 15 |
|
|
|
progress(1.0, "Preparing results") |
|
results = f"Number of persons/clusters detected: {num_clusters}\n\n" |
|
results += f"Breakdown of persons/clusters:\n" |
|
for cluster_id in range(num_clusters): |
|
results += f"Person/Cluster {cluster_id + 1}: {len([c for c in clusters if c == cluster_id])} frames\n" |
|
|
|
end_time = time.time() |
|
execution_time = end_time - start_time |
|
|
|
|
|
anomaly_faces_embeddings = [ |
|
cv2.imread(os.path.join(aligned_faces_folder, f"frame_{frame}_face.jpg")) |
|
for frame in anomaly_frames_embeddings |
|
if os.path.exists(os.path.join(aligned_faces_folder, f"frame_{frame}_face.jpg")) |
|
] |
|
anomaly_faces_embeddings = [cv2.cvtColor(face, cv2.COLOR_BGR2RGB) for face in anomaly_faces_embeddings if face is not None] |
|
|
|
return ( |
|
execution_time, |
|
results, |
|
df, |
|
mse_embeddings, |
|
mse_emotions, |
|
mse_plot_embeddings, |
|
mse_histogram_embeddings, |
|
*emotion_plots, |
|
face_samples["most_frequent"], |
|
face_samples["others"], |
|
anomaly_faces_embeddings, |
|
aligned_faces_folder |
|
) |
|
|
|
with gr.Blocks() as iface: |
|
gr.Markdown("# Facial Expressions Anomaly Detection") |
|
|
|
with gr.Row(): |
|
video_input = gr.Video() |
|
anomaly_threshold = gr.Slider(minimum=1, maximum=7, step=0.1, value=4.5, label="Anomaly Detection Threshold") |
|
fps_slider = gr.Slider(minimum=10, maximum=20, step=5, value=20, label="Frames Per Second") |
|
|
|
process_btn = gr.Button("Process Video") |
|
|
|
execution_time = gr.Number(label="Execution Time (seconds)") |
|
results_text = gr.Textbox(label="Anomaly Detection Results") |
|
|
|
anomaly_frames_embeddings = gr.Gallery(label="Anomaly Frames (Facial Embeddings)", columns=6, rows=2, height="auto") |
|
|
|
mse_embeddings_plot = gr.Plot(label="MSE: Facial Embeddings") |
|
mse_embeddings_hist = gr.Plot(label="MSE Distribution: Facial Embeddings") |
|
|
|
|
|
emotion_plots = [gr.Plot(label=f"{emotion.capitalize()} Over Time") for emotion in ['angry', 'disgust', 'fear', 'sad', 'happy']] |
|
|
|
face_samples_most_frequent = gr.Gallery(label="Most Frequent Person Samples (Target)", columns=6, rows=2, height="auto") |
|
face_samples_others = gr.Gallery(label="Other Persons Samples", columns=6, rows=1, height="auto") |
|
|
|
|
|
df_store = gr.State() |
|
mse_emotions_store = gr.State() |
|
mse_embeddings_store = gr.State() |
|
aligned_faces_folder_store = gr.State() |
|
|
|
process_btn.click( |
|
process_video, |
|
inputs=[video_input, anomaly_threshold, fps_slider], |
|
outputs=[ |
|
execution_time, results_text, df_store, mse_embeddings_store, mse_emotions_store, |
|
mse_embeddings_plot, mse_embeddings_hist, |
|
*emotion_plots, |
|
face_samples_most_frequent, face_samples_others, anomaly_frames_embeddings, |
|
aligned_faces_folder_store |
|
] |
|
) |
|
|
|
if __name__ == "__main__": |
|
iface.launch() |