From 326c1a33da4d8014f6c935243ec615814132adf4 Mon Sep 17 00:00:00 2001 From: Dillon Laird Date: Mon, 9 Sep 2024 14:34:41 -0700 Subject: [PATCH] fix edge case for OCR --- vision_agent/tools/tools.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/vision_agent/tools/tools.py b/vision_agent/tools/tools.py index d1734915..ef150f55 100644 --- a/vision_agent/tools/tools.py +++ b/vision_agent/tools/tools.py @@ -468,7 +468,7 @@ def ocr(image: np.ndarray) -> List[Dict[str, Any]]: pil_image = Image.fromarray(image).convert("RGB") image_size = pil_image.size[::-1] - if image_size[0] < 1 and image_size[1] < 1: + if image_size[0] < 1 or image_size[1] < 1: return [] image_buffer = io.BytesIO() pil_image.save(image_buffer, format="PNG") @@ -1112,6 +1112,8 @@ def florence2_ocr(image: np.ndarray) -> List[Dict[str, Any]]: """ image_size = image.shape[:2] + if image_size[0] < 1 or image_size[1] < 1: + return [] image_b64 = convert_to_b64(image) data = { "image": image_b64,