app.py

import streamlit as st
import cv2
import numpy as np
import tempfile
import os
from pathlib import Path
from typing import Optional, Tuple
from moviepy.editor import VideoFileClip
import torch
from PIL import Image

# ==============================
# Streamlit page config & Custom CSS
# ==============================
st.set_page_config(
   page_title="Ansim Blur - Face Privacy Protection",
   page_icon="🔒",
   layout="wide",
   initial_sidebar_state="expanded"
)

# Custom CSS - 레이아웃 안정화를 위한 수정
st.markdown("""
<style>
   /* 컨테이너 고정 높이 설정으로 흔들림 방지 */
   .image-container {
       min-height: 400px;
       display: flex;
       align-items: center;
       justify-content: center;
   }
   
   /* 메인 컨테이너 안정화 */
   .main .block-container {
       max-width: 1400px;
       padding-top: 2rem;
       padding-bottom: 2rem;
   }
   
   /* 컬럼 고정 */
   [data-testid="column"] {
       min-height: 500px;
   }
   
   /* 이미지 업로더 영역 고정 */
   [data-testid="stFileUploader"] {
       min-height: 150px;
   }
   
   /* 버튼 영역 고정 */
   .stButton {
       min-height: 60px;
   }
   
   /* 프로그레스 바 영역 고정 */
   .stProgress {
       min-height: 30px;
   }
   
   /* 헤더 스타일링 */
   h1 {
       background: linear-gradient(120deg, #a855f7, #ec4899);
       -webkit-background-clip: text;
       -webkit-text-fill-color: transparent;
       font-size: 3rem !important;
       font-weight: 700 !important;
       text-align: center;
       margin-bottom: 1rem !important;
   }
   
   /* 카드 스타일 */
   .stat-card {
       background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
       padding: 1rem;
       border-radius: 12px;
       color: white;
       text-align: center;
       height: 100px;
       display: flex;
       flex-direction: column;
       justify-content: center;
       box-shadow: 0 4px 15px rgba(0,0,0,0.1);
   }
   
   .stat-number {
       font-size: 2rem;
       margin-bottom: 0.3rem;
   }
   
   .stat-label {
       font-size: 0.85rem;
       opacity: 0.95;
   }
   
   /* 버튼 스타일 개선 */
   .stButton > button {
       background: linear-gradient(135deg, #a855f7 0%, #ec4899 100%);
       color: white;
       border: none;
       padding: 0.7rem 1.5rem;
       font-size: 1rem;
       font-weight: 600;
       border-radius: 25px;
       width: 100%;
       transition: transform 0.2s;
   }
   
   .stButton > button:hover {
       transform: translateY(-2px);
   }
   
   /* 사이드바 스타일 */
   .css-1d391kg {
       background-color: #f8f7ff;
   }
   
   /* Info 박스 */
   .info-box {
       background: #f0f4ff;
       border-left: 4px solid #667eea;
       padding: 1rem;
       border-radius: 8px;
       margin: 1rem 0;
   }
</style>
""", unsafe_allow_html=True)

# ==============================
# Header Section
# ==============================
st.markdown("<h1>🔒 Ansim Blur</h1>", unsafe_allow_html=True)
st.markdown("<p style='text-align: center; color: #6b7280; margin-bottom: 1rem;'>Advanced Face Privacy Protection</p>", unsafe_allow_html=True)

# Discord 배지를 가운데 정렬
st.markdown("""
<div style='text-align: center; margin-bottom: 2rem;'>
   <a href="https://discord.gg/openfreeai" target="_blank">
       <img src="https://img.shields.io/static/v1?label=Discord&message=Openfree%20AI&color=%230000ff&labelColor=%23800080&logo=discord&logoColor=white&style=for-the-badge" alt="Discord badge">
   </a>
</div>
""", unsafe_allow_html=True)

# Stats 카드 - 컨테이너로 고정
stats_container = st.container()
with stats_container:
   col1, col2, col3, col4 = st.columns(4)
   with col1:
       st.markdown("""<div class='stat-card'><div class='stat-number'>🖼️</div><div class='stat-label'>Image Support</div></div>""", unsafe_allow_html=True)
   with col2:
       st.markdown("""<div class='stat-card'><div class='stat-number'>🎥</div><div class='stat-label'>Video Processing</div></div>""", unsafe_allow_html=True)
   with col3:
       st.markdown("""<div class='stat-card'><div class='stat-number'>⚡</div><div class='stat-label'>Real-time</div></div>""", unsafe_allow_html=True)
   with col4:
       st.markdown("""<div class='stat-card'><div class='stat-number'>🛡️</div><div class='stat-label'>Privacy First</div></div>""", unsafe_allow_html=True)

st.markdown("---")

# ==============================
# Model loader
# ==============================
@st.cache_resource(show_spinner=False)
def load_model(model_path: str = "yolov8-face-hf.pt", device: Optional[str] = None):
   from ultralytics import YOLO
   if device is None:
       if torch.cuda.is_available():
           device = "cuda"
       elif torch.backends.mps.is_available():
           device = "mps"
       else:
           device = "cpu"
   model = YOLO(model_path)
   model.to(device)
   return model, device

with st.spinner("Loading AI model..."):
   model, device = load_model()

# ==============================
# Sidebar - 고정된 설정
# ==============================
with st.sidebar:
   st.markdown("## ⚙️ Configuration")
   st.info(f"Device: **{device.upper()}**")
   
   st.markdown("### Detection Settings")
   conf = st.slider("Confidence Threshold", 0.05, 0.9, 0.25, 0.01)
   iou = st.slider("NMS IoU", 0.1, 0.9, 0.45, 0.01)
   expand_ratio = st.slider("Box Expansion", 0.0, 0.5, 0.05, 0.01)
   
   st.markdown("### Blur Settings")
   mode_choice = st.selectbox("Style", ["Gaussian Blur", "Mosaic Effect"])
   
   if mode_choice == "Gaussian Blur":
       blur_kernel = st.slider("Blur Intensity", 15, 151, 51, 2)
       mosaic = 15
   else:
       mosaic = st.slider("Mosaic Size", 5, 40, 15, 1)
       blur_kernel = 51
   
   use_half = st.checkbox("Half Precision (CUDA)", value=False)

# ==============================
# Helper functions
# ==============================
def _ensure_odd(x: int) -> int:
   return x if x % 2 == 1 else x + 1

def _choose_writer_size(w: int, h: int) -> Tuple[int, int]:
   return (w if w % 2 == 0 else w - 1, h if h % 2 == 0 else h - 1)

def _apply_anonymization(face_roi: np.ndarray, mode: str, blur_kernel: int, mosaic: int = 15) -> np.ndarray:
   if face_roi.size == 0:
       return face_roi
   if mode == "Gaussian Blur":
       k = _ensure_odd(max(blur_kernel, 15))
       return cv2.GaussianBlur(face_roi, (k, k), 0)
   else:
       m = max(2, mosaic)
       h, w = face_roi.shape[:2]
       face_small = cv2.resize(face_roi, (max(1, w // m), max(1, h // m)), interpolation=cv2.INTER_LINEAR)
       return cv2.resize(face_small, (w, h), interpolation=cv2.INTER_NEAREST)

def blur_faces_image(image_bgr, conf, iou, expand_ratio, mode, blur_kernel, mosaic, use_half):
   h, w = image_bgr.shape[:2]
   face_count = 0

   with torch.no_grad():
       if use_half and device == "cuda":
           torch.set_default_dtype(torch.float16)
       results = model.predict(image_bgr, conf=conf, iou=iou, verbose=False, device=device)
       if use_half and device == "cuda":
           torch.set_default_dtype(torch.float32)

   for r in results:
       boxes = r.boxes.xyxy.cpu().numpy() if hasattr(r.boxes, "xyxy") else []
       face_count = len(boxes)
       for x1, y1, x2, y2 in boxes:
           x1, y1, x2, y2 = map(int, [x1, y1, x2, y2])

           if expand_ratio > 0:
               bw = x2 - x1
               bh = y2 - y1
               dx = int(bw * expand_ratio)
               dy = int(bh * expand_ratio)
               x1 -= dx; y1 -= dy; x2 += dx; y2 += dy

           x1 = max(0, min(w, x1))
           x2 = max(0, min(w, x2))
           y1 = max(0, min(h, y1))
           y2 = max(0, min(h, y2))
           if x2 <= x1 or y2 <= y1:
               continue

           roi = image_bgr[y1:y2, x1:x2]
           image_bgr[y1:y2, x1:x2] = _apply_anonymization(roi, mode, blur_kernel, mosaic)

   return image_bgr, face_count

def blur_faces_video(input_path, output_path, conf, iou, expand_ratio, mode, blur_kernel, mosaic, update_callback, use_half):
   cap = cv2.VideoCapture(input_path)
   if not cap.isOpened():
       raise IOError("Cannot open video")

   in_w = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
   in_h = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
   fps = cap.get(cv2.CAP_PROP_FPS) or 25.0
   frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) or 0

   out_w, out_h = _choose_writer_size(in_w, in_h)
   fourcc = cv2.VideoWriter_fourcc(*"mp4v")
   temp_video_path = str(Path(output_path).with_name("blurred_temp_video.mp4"))
   out = cv2.VideoWriter(temp_video_path, fourcc, fps, (out_w, out_h))

   idx = 0
   total_faces = 0
   try:
       while True:
           ret, frame = cap.read()
           if not ret:
               break
           frame = cv2.resize(frame, (out_w, out_h))

           with torch.no_grad():
               if use_half and device == "cuda":
                   torch.set_default_dtype(torch.float16)
               results = model.predict(frame, conf=conf, iou=iou, verbose=False, device=device)
               if use_half and device == "cuda":
                   torch.set_default_dtype(torch.float32)

           h, w = frame.shape[:2]
           r0 = results[0] if len(results) else None
           boxes = r0.boxes.xyxy if (r0 and hasattr(r0, "boxes")) else []
           total_faces += len(boxes)

           for b in boxes:
               x1, y1, x2, y2 = map(int, b)
               if expand_ratio > 0:
                   bw = x2 - x1
                   bh = y2 - y1
                   dx = int(bw * expand_ratio)
                   dy = int(bh * expand_ratio)
                   x1 -= dx; y1 -= dy; x2 += dx; y2 += dy

               x1 = max(0, min(w, x1))
               x2 = max(0, min(w, x2))
               y1 = max(0, min(h, y1))
               y2 = max(0, min(h, y2))
               if x2 <= x1 or y2 <= y1:
                   continue

               roi = frame[y1:y2, x1:x2]
               frame[y1:y2, x1:x2] = _apply_anonymization(roi, mode, blur_kernel, mosaic)

           out.write(frame)
           idx += 1
           if update_callback and frames > 0:
               update_callback(min(0.98, idx / frames), idx, frames, total_faces)
   finally:
       cap.release()
       out.release()

   try:
       if update_callback:
           update_callback(0.99, idx, frames, total_faces)
       original = VideoFileClip(input_path)
       processed = VideoFileClip(temp_video_path).set_audio(original.audio)
       processed.write_videofile(
           output_path,
           codec="libx264",
           audio_codec="aac",
           threads=1,
           logger=None
       )
       if update_callback:
           update_callback(1.0, idx, frames, total_faces)
       return output_path, total_faces
   except Exception as e:
       print("Audio merging failed:", e)
       return temp_video_path, total_faces

# ==============================
# Main Interface - 고정된 레이아웃
# ==============================
tab1, tab2 = st.tabs(["📸 Image Processing", "🎬 Video Processing"])

with tab1:
   # 고정된 컨테이너 생성
   main_container = st.container()
   
   with main_container:
       # 2개의 고정된 컬럼
       col1, col2 = st.columns(2, gap="large")
       
       # 왼쪽 컬럼 - 입력
       with col1:
           st.markdown("### Input")
           
           # 파일 업로더 컨테이너
           upload_container = st.container()
           with upload_container:
               uploaded_file = st.file_uploader(
                   "Choose an image",
                   type=["jpg", "png", "jpeg"],
                   key="img_upload"
               )
           
           # 원본 이미지 표시 영역 (고정 높이)
           original_placeholder = st.empty()
           info_placeholder = st.empty()
           
           if uploaded_file:
               file_bytes = np.asarray(bytearray(uploaded_file.read()), dtype=np.uint8)
               image = cv2.imdecode(file_bytes, cv2.IMREAD_COLOR)
               original_placeholder.image(cv2.cvtColor(image, cv2.COLOR_BGR2RGB), caption="Original", use_container_width=True)
               h, w = image.shape[:2]
               info_placeholder.info(f"Size: {w} × {h} pixels")
           else:
               # 빈 공간 유지
               original_placeholder.markdown("<div class='image-container'><p style='text-align:center;color:#999;'>No image uploaded</p></div>", unsafe_allow_html=True)
               info_placeholder.empty()
       
       # 오른쪽 컬럼 - 결과
       with col2:
           st.markdown("### Result")
           
           # 버튼 컨테이너
           button_container = st.container()
           
           # 결과 이미지 표시 영역 (고정 높이)
           result_placeholder = st.empty()
           success_placeholder = st.empty()
           download_placeholder = st.empty()
           
           with button_container:
               if uploaded_file:
                   if st.button("🔍 Process Image", type="primary", use_container_width=True):
                       with st.spinner("Processing..."):
                           result, face_count = blur_faces_image(
                               image.copy(), conf, iou, expand_ratio,
                               mode_choice, blur_kernel, mosaic, use_half
                           )
                           
                           result_placeholder.image(cv2.cvtColor(result, cv2.COLOR_BGR2RGB), caption="Processed", use_container_width=True)
                           success_placeholder.success(f"Blurred {face_count} face(s)")
                           
                           _, buffer = cv2.imencode('.jpg', result)
                           download_placeholder.download_button(
                               "⬇️ Download",
                               data=buffer.tobytes(),
                               file_name="blurred.jpg",
                               mime="image/jpeg",
                               use_container_width=True
                           )
               else:
                   result_placeholder.markdown("<div class='image-container'><p style='text-align:center;color:#999;'>Results will appear here</p></div>", unsafe_allow_html=True)

with tab2:
   video_container = st.container()
   
   with video_container:
       col1, col2 = st.columns(2, gap="large")
       
       with col1:
           st.markdown("### Input Video")
           
           video_upload = st.file_uploader(
               "Choose a video",
               type=["mp4", "avi", "mov", "mkv"],
               key="video_upload"
           )
           
           video_placeholder = st.empty()
           
           if video_upload:
               video_placeholder.video(video_upload)
           else:
               video_placeholder.markdown("<div class='image-container'><p style='text-align:center;color:#999;'>No video uploaded</p></div>", unsafe_allow_html=True)
       
       with col2:
           st.markdown("### Processed Video")
           
           process_button = st.empty()
           progress_placeholder = st.empty()
           stats_placeholder = st.empty()
           result_video_placeholder = st.empty()
           download_video_placeholder = st.empty()
           
           if video_upload:
               if process_button.button("🎬 Process Video", type="primary", use_container_width=True):
                   # Save uploaded file
                   input_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
                   with open(input_path, "wb") as f:
                       f.write(video_upload.read())
                   
                   output_path = str(Path(tempfile.gettempdir()) / "blurred_video.mp4")
                   
                   def update_progress(value, current_frame=0, total_frames=0, faces=0):
                       percent = int(value * 100)
                       progress_placeholder.progress(value)
                       if total_frames > 0:
                           stats_placeholder.info(f"📊 Frame: {current_frame}/{total_frames} | Progress: {percent}% | Faces: {faces}")
                   
                   try:
                       final_output, total_faces = blur_faces_video(
                           input_path, output_path,
                           conf=conf, iou=iou, expand_ratio=expand_ratio,
                           mode=mode_choice, blur_kernel=blur_kernel,
                           mosaic=mosaic,
                           update_callback=update_progress, use_half=use_half
                       )
                       
                       stats_placeholder.success(f"✅ Complete! Blurred {total_faces} faces.")
                       result_video_placeholder.video(final_output)
                       
                       with open(final_output, "rb") as file:
                           download_video_placeholder.download_button(
                               "⬇️ Download Video",
                               file,
                               file_name="blurred_video.mp4",
                               mime="video/mp4",
                               use_container_width=True
                           )
                   except Exception as e:
                       stats_placeholder.error(f"❌ Error: {e}")
                   finally:
                       if os.path.exists(input_path):
                           os.remove(input_path)
           else:
               result_video_placeholder.markdown("<div class='image-container'><p style='text-align:center;color:#999;'>Processed video will appear here</p></div>", unsafe_allow_html=True)