Merge branch 'rednote-hilab:master' into master

demo/demo_gradio_annotion.py

@@ -260,7 +260,7 @@ def process_image_inference_with_annotation(annotation_data, test_image_input,
     if image is None:
         return None, "Please select a test image or add an image in the annotation component", "", "", gr.update(value=None), ""
     if bbox is None:
-        return "Please select a bounding box by mouse", "Please select a bounding box by mouse", "", "", gr.update(value=None)
+        return None, "Please select a bounding box by mouse", "Please select a bounding box by mouse", "", "", gr.update(value=None)
     
     try:
         # Process using DotsOCRParser, passing the bbox parameter

dots_ocr/model/inference.py

@@ -1,8 +1,3 @@
-import json
-import io
-import base64
-import math
-from PIL import Image
 import requests
 from dots_ocr.utils.image_utils import PILimage_to_base64
 from openai import OpenAI

dots_ocr/parser.py

@@ -156,7 +156,7 @@ class DotsOCRParser:
             min_pixels = min_pixels or MIN_PIXELS  # preprocess image to the final input
             max_pixels = max_pixels or MAX_PIXELS
         if min_pixels is not None: assert min_pixels >= MIN_PIXELS, f"min_pixels should >= {MIN_PIXELS}"
-        if max_pixels is not None: assert max_pixels <= MAX_PIXELS, f"max_pixels should <+ {MAX_PIXELS}"
+        if max_pixels is not None: assert max_pixels <= MAX_PIXELS, f"max_pixels should <= {MAX_PIXELS}"
 
         if source == 'image' and fitz_preprocess:
             image = get_image_by_fitz_doc(origin_image, target_dpi=self.dpi)