Update emotion.py

emotion.py

@@ -1,5 +1,5 @@
 import time
-import cv2  # 確保有安裝 opencv-python
+import cv2
 from dataclasses import dataclass, field
 from typing import Optional, Dict, Tuple, List
 import gradio as gr
@@ -8,7 +8,7 @@ from PIL import Image
 from deepface import DeepFace
 
 # =========================
-# 參數（可微調）
+# 參數（效能優化版）
 # =========================
 ANALYZE_EVERY_SEC = 0.8
 DOWNSAMPLE_W = 720
@@ -16,7 +16,7 @@ EMA_ALPHA = 0.45
 CONF_TH = 40.0
 SWITCH_CONFIRM_SEC = 2.5
 IOU_TH = 0.25
-ZOOM_FACTOR = 1.3  # 微調縮放，讓使用者較容易對準
+ZOOM_FACTOR = 1.3
 DETECTOR_BACKEND = "mtcnn"
 ALIGN_FACE = True
 
@@ -25,8 +25,8 @@ ALIGN_FACE = True
 # =========================
 EMO_ZH = {"angry": "生氣", "disgust": "厭惡", "fear": "害怕", "happy": "開心", "sad": "難過", "surprise": "驚訝", "neutral": "平靜", "unknown": "未知"}
 EMO_TO_SCORE = {"angry": 1, "disgust": 1, "fear": 1, "sad": 2, "neutral": 3, "surprise": 4, "happy": 5, "unknown": 3}
-SCORE_LABELS = {1: "1（心情差）", 2: "2（不太好）", 3: "3（普通）", 4: "4（不錯）", 5: "5（超棒）"}
-SCORE_CHOICES = [SCORE_LABELS[i] for i in [1, 2, 3, 4, 5]]
+SCORE_LABELS = {i: f"{i}（{desc}）" for i, desc in zip(range(1,6), ["心情差","不太好","普通","不錯","超棒"])}
+SCORE_CHOICES = [SCORE_LABELS[i] for i in range(1,6)]
 
 @dataclass
 class AppState:
@@ -34,6 +34,8 @@ class AppState:
     running: bool = True
     finished: bool = False
     last_analyze_t: float = 0.0
+    last_check_t: float = 0.0  # 新增：控制檢查位置的頻率
+    is_aligned: bool = False   # 新增：快取對準狀態
     track_bbox: Optional[Tuple[int, int, int, int]] = None
     ema_scores: Optional[np.ndarray] = None
     emo_keys: List[str] = field(default_factory=list)
@@ -45,203 +47,132 @@ class AppState:
     final_score: int = 3
 
 # =========================
-# 輔助函式
+# 介面 HTML
 # =========================
 def _hint_html(msg: str) -> str:
     return f'<div style="border-radius:14px;padding:14px;border:1px dashed #DADADA;background:#FAFAFA;color:#000000;">{msg}</div>'
 
-def _camera_hint_before_start() -> str:
-    return _hint_html("請將臉部對準螢幕中央的方框。")
-
-def _camera_hint_done() -> str:
-    return _hint_html("已完成辨識。")
-
 def _result_card_html(emo_key: str, conf: Optional[float]) -> str:
     zh = EMO_ZH.get(emo_key, emo_key)
     score = EMO_TO_SCORE.get(emo_key, 3)
     conf_txt = "" if conf is None else f"{conf:.1f}%"
-    score_desc = SCORE_LABELS[score].split("（")[1].rstrip("）")
-    return f"""
-    <div style="border-radius:16px;padding:20px;border:1px solid #E6E6E6;background:#FFFFFF;color:#000000;">
-      <div style="font-size:14px;margin-bottom:8px;">辨識結果</div>
-      <div style="font-size:42px;font-weight:800;line-height:1.1;margin-bottom:14px;">{zh}</div>
-      <div style="font-size:16px;margin-bottom:6px;">心情分數：{score}（{score_desc}）</div>
-      <div style="font-size:14px;">信心值：{conf_txt}</div>
-    </div>
-    """
-
-def _score_radio_value(score: int) -> str:
-    return SCORE_LABELS[min(5, max(1, int(score)))]
-
-def _zoom_center(frame: np.ndarray, zoom: float) -> np.ndarray:
-    if frame is None or zoom <= 1.0: return frame
-    h, w = frame.shape[:2]
-    new_w, new_h = int(w / zoom), int(h / zoom)
-    start_x, start_y = (w - new_w) // 2, (h - new_h) // 2
-    return frame[start_y : start_y + new_h, start_x : start_x + new_w]
+    return f'<div style="border-radius:16px;padding:20px;border:1px solid #E6E6E6;background:#FFFFFF;color:#000000;"><div style="font-size:14px;margin-bottom:8px;">辨識結果</div><div style="font-size:42px;font-weight:800;line-height:1.1;margin-bottom:14px;">{zh}</div><div style="font-size:16px;margin-bottom:6px;">心情分數：{score}</div><div style="font-size:14px;">信心值：{conf_txt}</div></div>'
 
-def _downsample_rgb(frame: np.ndarray, target_w: int) -> np.ndarray:
-    if frame is None: return None
+# =========================
+# 影像處理
+# =========================
+def _zoom_center(frame, zoom):
+    if frame is None or zoom <= 1.0: return frame
     h, w = frame.shape[:2]
-    if w <= target_w: return frame
-    scale = target_w / w
-    new_h = max(1, int(h * scale))
-    return np.array(Image.fromarray(frame).resize((target_w, new_h)))
+    nw, nh = int(w / zoom), int(h / zoom)
+    sx, sy = (w - nw) // 2, (h - nh) // 2
+    return frame[sy:sy+nh, sx:sx+nw]
 
-# 核心修改：繪製固定導引框
 def _draw_ui_overlay(frame: np.ndarray, is_aligned: bool):
     canvas = frame.copy()
     h, w = canvas.shape[:2]
-    
-    # 1. 定義「固定導引框」的位置與大小 (置中)
-    bw, bh = int(w * 0.45), int(h * 0.55)  # 框的大小
+    bw, bh = int(w * 0.45), int(h * 0.55)
     x1, y1 = (w - bw) // 2, (h - bh) // 2
     x2, y2 = x1 + bw, y1 + bh
     
-    # 2. 顏色：對準了變綠色，沒對準是紅色
-    # (注意：OpenCV 預設是 BGR，但 Gradio 通常是 RGB，所以這裡設 (R, G, B))
     color = (0, 255, 0) if is_aligned else (255, 0, 0)
-    status_text = "SCANNING..." if is_aligned else "PLACE FACE HERE"
-
-    # 3. 繪製半透明遮罩 (讓框以外的地方變暗，更有提示感)
-    overlay = canvas.copy()
-    cv2.rectangle(overlay, (0, 0), (w, h), (0, 0, 0), -1)
-    alpha = 0.3 # 變暗程度
-    cv2.addWeighted(overlay, alpha, canvas, 1 - alpha, 0, canvas)
-    
-    # 4. 把框框的部分「挖亮」(恢復原樣)
-    canvas[y1:y2, x1:x2] = frame[y1:y2, x1:x2]
-
-    # 5. 繪製導引矩形
-    cv2.rectangle(canvas, (x1, y1), (x2, y2), color, 3)
-    
-    # 6. 繪製角點強化
-    l = 30 # 角線長度
-    cv2.line(canvas, (x1, y1), (x1+l, y1), color, 6)
-    cv2.line(canvas, (x1, y1), (x1, y1+l), color, 6)
-    cv2.line(canvas, (x2, y2), (x2-l, y2), color, 6)
-    cv2.line(canvas, (x2, y2), (x2, y2-l), color, 6)
-
-    # 7. 加上提示文字
-    cv2.putText(canvas, status_text, (x1, y1 - 15), cv2.FONT_HERSHEY_DUPLEX, 0.9, color, 2)
-    
+    text = "SCANNING..." if is_aligned else "PLACE FACE HERE"
+
+    # 效能優化：不再使用 addWeighted 製作全畫面遮罩，改用簡單線條
+    cv2.rectangle(canvas, (x1, y1), (x2, y2), color, 2)
+    l = 40
+    cv2.line(canvas, (x1, y1), (x1+l, y1), color, 5)
+    cv2.line(canvas, (x1, y1), (x1, y1+l), color, 5)
+    cv2.line(canvas, (x2, y2), (x2-l, y2), color, 5)
+    cv2.line(canvas, (x2, y2), (x2, y2-l), color, 5)
+    cv2.putText(canvas, text, (x1, y1 - 15), cv2.FONT_HERSHEY_SIMPLEX, 0.8, color, 2)
     return canvas
 
-def _is_face_in_guide(face_bbox, frame_shape):
-    """ 檢查人臉是否在中央引導框內 """
-    if face_bbox is None: return False
-    fx, fy, fw, fh = face_bbox
-    h, w = frame_shape[:2]
-    
-    # 引導框範圍 (需與 _draw_ui_overlay 一致)
-    bw, bh = int(w * 0.45), int(h * 0.55)
-    gx1, gy1 = (w - bw) // 2, (h - bh) // 2
-    gx2, gy2 = gx1 + bw, gy1 + bh
-    
-    # 計算人臉中心點
-    fcx, fcy = fx + fw // 2, fy + fh // 2
-    
-    # 判定標準：臉部中心點在引導框內
-    return (gx1 < fcx < gx2) and (gy1 < fcy < gy2)
-
-def _pick_face_by_tracking(faces, prev_bbox):
-    if not faces: return None, None
-    cand = []
-    for f in faces:
-        fa = f.get("facial_area", {})
-        x, y, w, h = fa.get("x",0), fa.get("y",0), fa.get("w",0), fa.get("h",0)
-        if w > 0 and h > 0: cand.append((f, (x, y, w, h), w * h))
-    if not cand: return None, None
-    return sorted(cand, key=lambda x: x[2], reverse=True)[0][:2]
-
 # =========================
-# Camera streaming
+# 串流核心
 # =========================
 def on_stream(frame_rgb: np.ndarray, st: AppState):
     if frame_rgb is None or st.finished or (not st.running):
         return (None, gr.update(), gr.update(), gr.update(), st, gr.update())
 
-    # 1. 數位變焦
-    frame_rgb = _zoom_center(frame_rgb, ZOOM_FACTOR)
     now = time.time()
-    
-    # 2. 快速人臉檢測 (判斷位置)
-    small_check = _downsample_rgb(frame_rgb, 400)
-    faces_check = DeepFace.extract_faces(img_path=small_check, detector_backend="opencv", enforce_detection=False)
-    pick_check, bbox_check = _pick_face_by_tracking(faces_check, st.track_bbox)
-    
-    # 3. 檢查人臉是否對準中央框
-    is_aligned = _is_face_in_guide(bbox_check, frame_rgb.shape)
-    
-    # 4. 繪製導引畫面 (紅/綠框與遮罩)
-    frame_with_ui = _draw_ui_overlay(frame_rgb, is_aligned)
+    frame_rgb = _zoom_center(frame_rgb, ZOOM_FACTOR)
 
-    # 5. 分析邏輯 (只有對準時才進行深度分析)
-    if is_aligned and (now - st.last_analyze_t >= ANALYZE_EVERY_SEC):
+    # --- 效能優化點：降低檢查頻率 (每 0.3 秒才檢查一次人臉位置) ---
+    if now - st.last_check_t > 0.3:
+        st.last_check_t = now
+        # 極小化影像以利快速檢查
+        small_check = np.array(Image.fromarray(frame_rgb).resize((250, 250)))
+        try:
+            faces = DeepFace.extract_faces(img_path=small_check, detector_backend="opencv", enforce_detection=False)
+            if faces and faces[0]['face'] is not None:
+                fa = faces[0]['facial_area']
+                # 將座標還原到原圖比例
+                fx, fy = fa['x'] * (frame_rgb.shape[1]/250), fa['y'] * (frame_rgb.shape[0]/250)
+                fw, fh = fa['w'] * (frame_rgb.shape[1]/250), fa['h'] * (frame_rgb.shape[0]/250)
+                
+                # 檢查中心點是否在導引框內 (中央 45%x55%)
+                h, w = frame_rgb.shape[:2]
+                gx1, gy1 = (w - int(w*0.45))//2, (h - int(h*0.55))//2
+                gx2, gy2 = gx1 + int(w*0.45), gy1 + int(h*0.55)
+                st.is_aligned = (gx1 < fx + fw/2 < gx2) and (gy1 < fy + fh/2 < gy2)
+            else:
+                st.is_aligned = False
+        except:
+            st.is_aligned = False
+
+    # 繪製畫面 (這個步驟很快)
+    frame_with_ui = _draw_ui_overlay(frame_rgb, st.is_aligned)
+
+    # --- 深度分析邏輯 (只有對準時執行) ---
+    if st.is_aligned and (now - st.last_analyze_t >= ANALYZE_EVERY_SEC):
         st.last_analyze_t = now
-        small = _downsample_rgb(frame_rgb, DOWNSAMPLE_W)
-        faces = DeepFace.extract_faces(img_path=small, detector_backend=DETECTOR_BACKEND, enforce_detection=False, align=ALIGN_FACE)
-        pick, bbox = _pick_face_by_tracking(faces, st.track_bbox)
+        # 進行高品質偵測與分析
+        res = DeepFace.analyze(img_path=frame_rgb, actions=["emotion"], enforce_detection=False, detector_backend="opencv")
+        emo_dict = res[0].get("emotion") if isinstance(res, list) else res.get("emotion")
         
-        if pick:
-            st.track_bbox = bbox
-            face = pick["face"]
-            if face.dtype != np.uint8: face = (face * 255).astype(np.uint8)
+        if emo_dict:
+            top_key = max(emo_dict, key=lambda k: float(emo_dict[k]))
+            keys = list(emo_dict.keys())
+            vec = np.array([float(emo_dict[k]) for k in keys])
+            if st.ema_scores is None: st.ema_scores, st.emo_keys = vec, keys
+            else: st.ema_scores = EMA_ALPHA * vec + (1.0 - EMA_ALPHA) * st.ema_scores
             
-            res = DeepFace.analyze(img_path=face, actions=["emotion"], enforce_detection=False)
-            emo_dict = res[0].get("emotion") if isinstance(res, list) else res.get("emotion")
+            stable_key = st.emo_keys[np.argmax(st.ema_scores)]
+            st.stable_emo = stable_key
             
-            if emo_dict:
-                top_key = max(emo_dict, key=lambda k: float(emo_dict[k]))
-                keys = list(emo_dict.keys())
-                vec = np.array([float(emo_dict[k]) for k in keys])
-                if st.ema_scores is None: st.ema_scores, st.emo_keys = vec, keys
-                else: st.ema_scores = EMA_ALPHA * vec + (1.0 - EMA_ALPHA) * st.ema_scores
-                
-                stable_key = st.emo_keys[np.argmax(st.ema_scores)]
-                st.stable_emo = stable_key
-                
-                if st.candidate_emo != stable_key:
-                    st.candidate_emo, st.candidate_since = stable_key, now
-                elif st.candidate_since and (now - st.candidate_since >= SWITCH_CONFIRM_SEC):
-                    st.final_emo, st.final_conf, st.running, st.finished = stable_key, float(emo_dict[top_key]), False, True
-                    return (frame_with_ui, _result_card_html(stable_key, st.final_conf), _score_radio_value(EMO_TO_SCORE.get(stable_key, 3)), _camera_hint_done(), st, gr.update(value="重新攝影機辨識"))
-
-    # 顯示穩定情緒
-    show_emo = st.stable_emo or "unknown"
-    return (
-        frame_with_ui, 
-        gr.update(value=_result_card_html(show_emo, st.final_conf)), 
-        gr.update(value=_score_radio_value(EMO_TO_SCORE.get(show_emo, 3))), 
-        gr.update(value=_camera_hint_before_start()), 
-        st, 
-        gr.update()
-    )
+            if st.candidate_emo != stable_key:
+                st.candidate_emo, st.candidate_since = stable_key, now
+            elif st.candidate_since and (now - st.candidate_since >= SWITCH_CONFIRM_SEC):
+                st.final_emo, st.final_conf, st.running, st.finished = stable_key, float(emo_dict[top_key]), False, True
+                return (frame_with_ui, _result_card_html(stable_key, st.final_conf), SCORE_LABELS[EMO_TO_SCORE.get(stable_key, 3)], "辨識完成", st, gr.update(value="重新攝影機辨識"))
+
+    return (frame_with_ui, gr.update(value=_result_card_html(st.stable_emo, st.final_conf)), gr.update(), gr.update(), st, gr.update())
 
+# =========================
+# 其他 Button 函式 (簡化版以維持穩定)
+# =========================
 def on_restart(st: AppState):
-    st.running, st.finished, st.last_analyze_t = True, False, 0.0
-    st.track_bbox, st.ema_scores, st.stable_emo, st.final_emo = None, None, "unknown", None
-    return (gr.update(visible=True), _hint_html("等待影像..."), _score_radio_value(3), _camera_hint_before_start(), st, "重新攝影機辨識")
+    st.running, st.finished, st.ema_scores, st.stable_emo, st.final_emo = True, False, None, "unknown", None
+    return (gr.update(visible=True), _hint_html("重新開始..."), SCORE_LABELS[3], "請對準方框", st, "重新攝影機辨識")
 
 def on_stop(st: AppState):
     st.running, st.finished = False, True
-    return (gr.update(visible=False), _hint_html("已停止"), _score_radio_value(3), _camera_hint_stopped(), st, "重新攝影機辨識")
+    return (gr.update(visible=False), _hint_html("已停止"), SCORE_LABELS[3], "停止", st, "重新攝影機辨識")
 
 # =========================
 # UI
 # =========================
 with gr.Blocks(title="Emotion Detector") as demo:
     state = gr.State(AppState())
-    gr.Markdown("## 情緒辨識系統")
-    hint = gr.HTML(_camera_hint_before_start())
+    gr.Markdown("## 情緒辨識系統 (效能優化版)")
+    hint = gr.HTML("請將臉部對準螢幕中央方框")
     cam = gr.Image(sources=["webcam"], streaming=True, type="numpy", label="攝影機預覽")
-    
     with gr.Row():
         btn_restart = gr.Button("重新攝影機辨識", variant="primary")
         btn_stop = gr.Button("停止", variant="secondary")
-        
     result = gr.HTML(_hint_html("等待串流..."))
-    mood = gr.Radio(choices=SCORE_CHOICES, value=SCORE_LABELS[3], label="系統判定心情", interactive=False)
+    mood = gr.Radio(choices=SCORE_CHOICES, value=SCORE_LABELS[3], label="判定心情", interactive=False)
 
     cam.stream(on_stream, [cam, state], [cam, result, mood, hint, state, btn_restart], show_progress="minimal")
     btn_restart.click(on_restart, [state], [cam, result, mood, hint, state, btn_restart])