FINAL FIX: Ultra-minimal Gradio interface to avoid schema bugs

- Strip down to bare minimum gr.Interface configuration
- Use string shortcuts ('image', 'text') instead of component objects
- Remove ALL complex parameters that trigger schema processing
- Simplify to single output text instead of dual outputs
- Eliminate any potential source of TypeError bool iteration
- This should definitively resolve the persistent schema error

app.py

@@ -1,113 +1 @@
-import os
-import cv2
-import numpy as np
-import gradio as gr
-
-# 檢查檔案是否存在
-model_path = "tsflow/models/best_model.pt"
-config_path = "tsflow/results/test_results.json"
-
-# 如果是在SignView2.0目錄下運行，調整路徑
-if not os.path.exists(model_path):
-    model_path = "../tsflow/models/best_model.pt"
-    config_path = "../tsflow/results/test_results.json"
-
-try:
-    from realtime_sign_prediction import RealtimeSignPredictor
-    
-    # 初始化預測器
-    print("🚀 正在初始化手語辨識系統...")
-    predictor = RealtimeSignPredictor(
-        model_path=model_path,
-        config_path=config_path,
-        sequence_length=50,
-        use_segmentation=True
-    )
-    print("✅ 手語辨識系統初始化完成！")
-    MODEL_LOADED = True
-    
-except Exception as e:
-    print(f"⚠️ 模型載入失敗: {e}")
-    print("🔄 使用模擬模式運行...")
-    MODEL_LOADED = False
-
-def process_image(image):
-    """處理上傳的影像"""
-    if image is None:
-        return None, "請上傳影像或使用攝像頭拍攝"
-    
-    if not MODEL_LOADED:
-        return image, "⚠️ 模型未載入，無法進行預測\n請檢查模型檔案路徑"
-    
-    try:
-        # 處理畫面
-        results, keypoints, flow_features = predictor.process_frame(image)
-        
-        # 繪製關鍵點
-        annotated_frame = predictor.draw_landmarks(image.copy(), results)
-        
-        # 獲取預測結果
-        top_predictions = predictor.get_top_predictions(top_k=3)
-        
-        # 格式化預測結果
-        if top_predictions:
-            prediction_text = "🎯 手語辨識結果:\n\n"
-            for i, (label, confidence) in enumerate(top_predictions, 1):
-                prediction_text += f"{i}. {label}: {confidence:.2%}\n"
-            
-            # 添加序列資訊
-            prediction_text += f"\n📊 序列進度: {len(predictor.keypoint_sequence)}/{predictor.sequence_length}"
-        else:
-            prediction_text = "📡 正在收集動作序列...\n請確保手語動作清晰可見"
-        
-        return annotated_frame, prediction_text
-        
-    except Exception as e:
-        return image, f"❌ 處理錯誤: {str(e)}"
-
-def clear_sequence():
-    """清除預測序列"""
-    if MODEL_LOADED:
-        predictor.keypoint_sequence.clear()
-        predictor.flow_sequence.clear()
-        return "✅ 已清除預測序列"
-    else:
-        return "⚠️ 模型未載入"
-
-# 使用最簡單且穩定的Gradio介面
-title = "🤟 SignView2.0 - 手語辨識系統"
-description = """
-**支援34種手語詞彙的即時辨識系統，準確率達94.25%**
-
-📋 **支援詞彙**: again, all, apple, bad, bathroom, beautiful, bird, black, blue, book, bored, boy, brother, brown, but, computer, cousin, dance, day, deaf, doctor, dog, draw, drink, eat, english, family, father, fine, finish, fish, forget, friend, girl
-
-🚀 **使用說明**: 上傳影像 → 點擊分析 → 查看結果
-
-📊 **系統資訊**: 準確率94.25% | F1分數94.24% | MediaPipe + 光流特徵 | BiLSTM + 注意力機制
-
-**開發者**: XiaoBai1221 | **平台**: Hugging Face Spaces
-"""
-
-# 使用Interface來避免Blocks的複雜問題
-demo = gr.Interface(
-    fn=process_image,
-    inputs=gr.Image(label="上傳手語影像"),
-    outputs=[
-        gr.Image(label="辨識結果"),
-        gr.Textbox(label="預測結果", lines=6)
-    ],
-    title=title,
-    description=description,
-    allow_flagging="never"
-)
-
-if __name__ == "__main__":
-    print("🎉 SignView2.0 手語辨識系統已啟動！")
-    
-    # 使用最簡化的launch參數，移除可能有問題的參數
-    demo.launch(
-        share=True,
-        server_name="0.0.0.0",
-        server_port=7860,
-        show_error=True
-    ) 
+