Spaces:
Sleeping
Sleeping
sotirios-slv
commited on
Commit
•
4286a1e
1
Parent(s):
cd373fc
Updated the excel export
Browse files
app.py
CHANGED
@@ -38,8 +38,8 @@ def run(image, lang="eng"):
|
|
38 |
print("Image ", image)
|
39 |
try:
|
40 |
print("Image filename ", image.filename)
|
41 |
-
except:
|
42 |
-
print("Could not print image filename")
|
43 |
result = pytesseract.image_to_string(image, lang=None if lang == [] else lang)
|
44 |
|
45 |
ner = get_named_entities(result)
|
@@ -58,14 +58,9 @@ def download_output(ocr_text: str, named_entities: str, image_name="test"):
|
|
58 |
ner_df = pd.Series(named_entities_list)
|
59 |
|
60 |
with pd.ExcelWriter(output_file) as writer:
|
61 |
-
ocr_df.to_excel(
|
62 |
-
writer, sheet_name="OCR text", columns=["OCR text"], index=False
|
63 |
-
)
|
64 |
ner_df.to_excel(
|
65 |
-
writer,
|
66 |
-
sheet_name="Named entities",
|
67 |
-
columns=["Named entities"],
|
68 |
-
index=False,
|
69 |
)
|
70 |
return output_file
|
71 |
|
|
|
38 |
print("Image ", image)
|
39 |
try:
|
40 |
print("Image filename ", image.filename)
|
41 |
+
except Exception as e:
|
42 |
+
print(f"Could not print image filename: {e}")
|
43 |
result = pytesseract.image_to_string(image, lang=None if lang == [] else lang)
|
44 |
|
45 |
ner = get_named_entities(result)
|
|
|
58 |
ner_df = pd.Series(named_entities_list)
|
59 |
|
60 |
with pd.ExcelWriter(output_file) as writer:
|
61 |
+
ocr_df.to_excel(writer, sheet_name="OCR text", columns=["OCR text"])
|
|
|
|
|
62 |
ner_df.to_excel(
|
63 |
+
writer, sheet_name="Named entities", columns=["Named entities"]
|
|
|
|
|
|
|
64 |
)
|
65 |
return output_file
|
66 |
|