update

tokenization_qwen.py

@@ -416,7 +416,7 @@ class QWenTokenizer(PreTrainedTokenizer):
         else:
             image = plt.imread(image)
             h, w = image.shape[0], image.shape[1]
-        visualizer = Visualizer(image)
+        visualizer = Visualizer(image, self.font_path)
 
         boxes = self._fetch_all_box_with_ref(response)
         if not boxes:
@@ -491,8 +491,9 @@ class VisImage:
 
 
 class Visualizer:
-    def __init__(self, img_rgb, metadata=None, scale=1.0):
+    def __init__(self, img_rgb, font_path, metadata=None, scale=1.0):
         self.img = np.asarray(img_rgb).clip(0, 255).astype(np.uint8)
+        self.font_path = os.path.join(font_path, "SimSun.ttf")
         self.output = VisImage(self.img, scale=scale)
         self.cpu_device = torch.device("cpu")
 
@@ -524,7 +525,7 @@ class Visualizer:
             y,
             text,
             size=font_size * self.output.scale,
-            fontproperties=FontProperties(fname=r"SimSun.ttf"),
+            fontproperties=FontProperties(fname=self.font_path),
             bbox={"facecolor": "black", "alpha": 0.8, "pad": 0.7, "edgecolor": "none"},
             verticalalignment="top",
             horizontalalignment=horizontal_alignment,