update OCR

requirements.txt

@@ -11,4 +11,4 @@ huggingface_hub
 timm
 surya-ocr
 vietocr
- 
+realesrgan

src/inference.py

@@ -28,6 +28,41 @@ _ocr_paddle_en = None
 _ocr_easyocr = None
 _ocr_vietocr = None
 
+
+# Real-ESRGAN upscaler (optional)
+REALESRGAN_AVAILABLE = False
+try:
+    from realesrgan import RealESRGANer
+    from basicsr.archs.rrdbnet_arch import RRDBNet
+    REALESRGAN_AVAILABLE = True
+    print("[INFO] Real-ESRGAN available")
+except ImportError:
+    print("[WARN] Real-ESRGAN not installed. Install: pip install realesrgan basicsr")
+
+def get_esrgan_upsampler():
+    if not REALESRGAN_AVAILABLE:
+        return None
+    model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=6, num_grow_ch=32, scale=4)
+    upsampler = RealESRGANer(
+        scale=4,
+        model_path='weights/RealESRGAN_x4plus_anime_6B.pth',
+        model=model,
+        device=DEVICE
+    )
+    return upsampler
+
+def upscale_if_needed(img_bgr, min_dim=300):
+    """Upscale image using Real-ESRGAN if both dimensions are below threshold."""
+    h, w = img_bgr.shape[:2]
+    if h < min_dim or w < min_dim:
+        upsampler = get_esrgan_upsampler()
+        if upsampler is not None:
+            try:
+                output, _ = upsampler.enhance(img_bgr, outscale=2)
+                return output
+            except Exception as e:
+                print(f"[WARN] ESRGAN upscale failed: {e}")
+    return img_bgr
 # ============================================================
 # DOMAIN DICTIONARY — Từ điển bản vẽ kỹ thuật Việt Nam
 # ============================================================
@@ -800,10 +835,10 @@ def detect_table_structure(img_bgr):
 # ============================================================
 
 def ocr_cell_improved(img_cell, backend="paddle"):
-    """OCR a single table cell with multiple attempts."""
     if img_cell is None or img_cell.size == 0:
         return ""
-    
+    # Upscale very small cells with ESRGAN
+    img_cell = upscale_if_needed(img_cell, min_dim=150)
     h, w = img_cell.shape[:2]
     if h < 5 or w < 5:
         return ""