Stability: Reverted to Gradio 3.50.2 with Premium CSS Styling

README.md

@@ -16,7 +16,7 @@ tags:
 datasets:
 - c3rl/IIIT-INDIC-HW-WORDS-Hindi
 sdk: gradio
-sdk_version: "4.44.1"
+sdk_version: 3.50.2
 python_version: "3.10"
 app_file: app.py
 pinned: true

app.py

@@ -10,19 +10,6 @@ from transformers import TrOCRProcessor, VisionEncoderDecoderModel
 from cnn_model import CharacterClassifier
 from preprocessing import preprocess_for_ocr
 
-# --- ROBUST GLOBAL PATCH FOR GRADIO 4.x ---
-import gradio_client.utils
-def robust_get_type(schema):
-    if isinstance(schema, bool): return "Any"
-    if not isinstance(schema, dict): return "Any"
-    if "const" in schema: return "Any"
-    return original_get_type(schema)
-
-if hasattr(gradio_client.utils, "get_type"):
-    original_get_type = gradio_client.utils.get_type
-    gradio_client.utils.get_type = robust_get_type
-# ------------------------------------------
-
 # --- CONFIGURATION ---
 BASE_MODEL_ID = "paudelanil/trocr-devanagari-2"
 ADAPTER_ID = "manishw10/devgen-trocr-devanagari-lora"
@@ -30,7 +17,7 @@ CNN_MODEL_PATH = "devanagari-cnn-classifier.pt"
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 # --- ENGINE INITIALIZATION ---
-print("System: Initializing Full Combined Suite...")
+print("System: Initializing Stable Premium Engine (3.50.2)...")
 processor = TrOCRProcessor.from_pretrained(BASE_MODEL_ID)
 base_model = VisionEncoderDecoderModel.from_pretrained(BASE_MODEL_ID)
 base_model.config.decoder_start_token_id = processor.tokenizer.cls_token_id
@@ -91,8 +78,8 @@ def get_confidence_html(confidence):
     return f"""<div style="display: flex; flex-direction: column; align-items: center; background: rgba(0,0,0,0.2); border-radius: 20px; padding: 15px;">
         <svg width="100" height="100" viewBox="0 0 100 100">
             <circle cx="50" cy="50" r="45" fill="none" stroke="rgba(255,255,255,0.1)" stroke-width="8" />
-            <circle cx="50" cy="50" r="45" fill="none" stroke="{color}" stroke-width="8" stroke-dasharray="282.7" stroke-dashoffset="{282.7 * (1 - confidence)}" stroke-linecap="round" style="transition: stroke-dashoffset 1s;" />
-            <text x="50" y="55" font-family="Outfit" font-size="20" font-weight="bold" fill="{color}" text-anchor="middle">{int(confidence * 100)}%</text>
+            <circle cx="50" cy="50" r="45" fill="none" stroke="{color}" stroke-width="8" stroke-dasharray="282.7" stroke-dashoffset="{282.7 * (1 - confidence)}" stroke-linecap="round" />
+            <text x="50" y="55" font-family="Arial" font-size="20" font-weight="bold" fill="{color}" text-anchor="middle">{int(confidence * 100)}%</text>
         </svg>
     </div>"""
 
@@ -103,9 +90,9 @@ def predict(image, manual_mode):
     pre_pil = preprocess_for_ocr(buf.getvalue())
     if manual_mode == "Automatic":
         mode, ar, bc = original_classify_input(pre_pil)
-        status = f"**System**: {mode.upper()} detected (AR: {ar:.2f}, Blobs: {bc})"
+        status = f"System: {mode.upper()} (AR: {ar:.2f}, Blobs: {bc})"
     else:
-        mode = manual_mode.lower(); status = f"**Manual Mode**: {mode.upper()}"
+        mode = manual_mode.lower(); status = f"Manual Mode: {mode.upper()}"
     try:
         if mode == "character" and cnn_engine.available:
             res = cnn_engine.predict(pre_pil)
@@ -120,30 +107,30 @@ def predict(image, manual_mode):
     except Exception as e:
         return pre_pil, f"Error: {str(e)}", "Failed", "None", ""
 
-# --- PREMIUM UI ---
+# --- PREMIUM CSS (Gradio 3.x Optimized) ---
 CSS = """
-@import url('https://fonts.googleapis.com/css2?family=Outfit:wght@400;600&family=Inter:wght@400;500&display=swap');
-.gradio-container { background: #0f172a !important; color: white !important; font-family: 'Inter', sans-serif !important; }
-.premium-card { background: rgba(30, 41, 59, 0.7) !important; backdrop-filter: blur(12px); border: 1px solid rgba(255,255,255,0.1); border-radius: 24px; padding: 2rem; box-shadow: 0 25px 50px -12px rgba(0,0,0,0.5); }
-.result-box { font-size: 3rem !important; font-weight: 600; text-align: center; color: #818cf8; background: transparent !important; border: none !important; }
+.gradio-container { background: linear-gradient(135deg, #0f172a 0%, #1e1b4b 100%) !important; color: white !important; }
+.premium-card { background: rgba(30, 41, 59, 0.7) !important; border: 1px solid rgba(255,255,255,0.1); border-radius: 20px; padding: 20px; box-shadow: 0 10px 30px rgba(0,0,0,0.5); }
+.result-box textarea { font-size: 2.5rem !important; font-weight: bold !important; color: #818cf8 !important; text-align: center !important; background: transparent !important; border: none !important; }
+h1 { color: #818cf8 !important; font-size: 2.5rem !important; }
 """
 
-with gr.Blocks(css=CSS, theme=gr.themes.Default()) as demo:
+with gr.Blocks(css=CSS) as demo:
     with gr.Column(elem_classes="premium-card"):
         gr.Markdown("# 🕉️ DevGen OCR")
         with gr.Row():
             with gr.Column(scale=1):
-                img_in = gr.Image(type="pil", label="Input Handwriting")
-                mode_ctrl = gr.Radio(["Automatic", "Word", "Character"], value="Automatic", label="Logic Mode")
+                img_in = gr.Image(type="pil", label="Input")
+                mode_ctrl = gr.Radio(["Automatic", "Word", "Character"], value="Automatic", label="Mode")
                 sub_btn = gr.Button("Recognize", variant="primary")
             with gr.Column(scale=1):
                 conf_html = gr.HTML()
-                text_out = gr.Textbox(label="Result", elem_classes="result-box", interactive=False, show_label=False)
-                status_md = gr.Markdown("Engine ready.")
-                engine_txt = gr.Textbox(label="Active Model", interactive=False)
+                text_out = gr.Textbox(label="Result", elem_classes="result-box", interactive=False)
+                status_md = gr.Markdown("Ready.")
+                engine_txt = gr.Textbox(label="Model", interactive=False)
         with gr.Column():
             gr.Markdown("### 🛠️ Visual Debug: What the Model Sees")
-            img_proc = gr.Image(type="pil", label="Preprocessed Input", interactive=False, show_label=False)
+            img_proc = gr.Image(type="pil", label="Preprocessed", interactive=False)
 
     sub_btn.click(predict, [img_in, mode_ctrl], [img_proc, text_out, status_md, engine_txt, conf_html])