docker-compose.yml

@@ -78,8 +78,8 @@ services:
       dockerfile: Dockerfile
     container_name: archive-ocr
     restart: unless-stopped
-    mem_limit: 6g
-    memswap_limit: 6g
+    mem_limit: 10g
+    memswap_limit: 10g
     volumes:
       - ocr_models:/app/models
     environment:

ocr-service/main.py

@@ -92,8 +92,9 @@ async def _download_and_convert_pdf(url: str) -> list[Image.Image]:
 
     for page_idx in range(len(pdf)):
         page = pdf[page_idx]
-        # Render at 300 DPI for good OCR quality
-        bitmap = page.render(scale=300 / 72)
+        # Render at 200 DPI — balances OCR quality vs memory usage
+        # (Surya 0.17 models use ~5GB idle; 300 DPI causes OOM on multi-page docs)
+        bitmap = page.render(scale=200 / 72)
         pil_image = bitmap.to_pil()
         images.append(pil_image)