pragnakalp commited on
Commit
51e91cb
·
1 Parent(s): 86d43bd

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +54 -51
app.py CHANGED
@@ -12,6 +12,7 @@ from huggingface_hub import Repository
12
  from datetime import datetime
13
  import scipy.ndimage.interpolation as inter
14
  import easyocr
 
15
  from PIL import Image
16
  from paddleocr import PaddleOCR
17
  import socket
@@ -20,30 +21,32 @@ from huggingface_hub import HfApi
20
  import smtplib
21
 
22
  HF_TOKEN = os.environ.get("HF_TOKEN")
23
- mydataset_name = "pragnakalp/OCR-img-to-text"
24
- print("$$$$$$$$$$$$$$$$$$$$$$$$$$$$",type(mydataset_name))
25
- hf_writer = gr.HuggingFaceDatasetSaver(HF_TOKEN,mydataset_name)
26
-
27
-
28
- # # DATASET_REPO_URL = "https://huggingface.co/datasets/pragnakalp/OCR-img-to-text"
29
- # DATA_FILENAME = "ocr_data.csv"
30
- # DATA_FILE = os.path.join("ocr_data", DATA_FILENAME)
31
- # HF_TOKEN = os.environ.get("HF_TOKEN")
32
- # DATASET_REPO_ID = "pragnakalp/OCR-img-to-text"
33
- # print("is none?", HF_TOKEN is None)
34
- # try:
35
- # hf_hub_download(
36
- # repo_id=DATASET_REPO_ID,
37
- # filename=DATA_FILENAME,
38
- # cache_dir=DATA_DIRNAME,
39
- # force_filename=DATA_FILENAME
40
- # )
41
- # except:
42
- # print("file not found")
43
-
44
- # repo = Repository(
45
- # local_dir="ocr_data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
46
- # )
 
 
47
 
48
  def get_device_ip_address():
49
 
@@ -128,31 +131,31 @@ def generate_ocr(Method,img):
128
  if Method == 'PaddleOCR':
129
  text_output = ocr_with_paddle(img)
130
 
131
- # with open(DATA_FILE, "a") as csvfile:
132
- # writer = csv.DictWriter(csvfile, fieldnames=["method", "image", "generated_text"])
133
- # writer.writerow(
134
- # {"method": Method, "image": img, "generated_text": text_output}
135
- # )
136
- # commit_url = repo.push_to_hub()
137
- # print(commit_url)
138
 
139
- # save_details(Method,text_output,img)
140
- # sender="pragnakalp.dev33@gmail.com"
141
- # password="httscgatatbbxxur"
142
- # reciever="pragnakalp.dev35@gmail.com"
143
-
144
- # s = smtplib.SMTP('smtp.gmail.com', 587)
145
- # s.starttls()
146
- # s.ehlo()
147
- # s.login(sender,password)
148
-
149
- # message = """Subject : Appointment Booking\n\n
150
- # Hello,
151
- # Your OCR generated successfully"""
152
- # s.sendmail(sender, reciever, message)
153
- # s.quit()
154
- # mailsend=1
155
- # print("Send mail successfully")
156
  return text_output
157
 
158
  except Exception as e:
@@ -220,8 +223,8 @@ demo = gr.Interface(
220
  title="Optical Character Recognition",
221
  description="Try OCR with different methods",
222
  css=".gradio-container {background-color: lightgray} #radio_div {background-color: #FFD8B4; font-size: 40px;}",
223
- allow_flagging = "manual",
224
- flagging_dir = "flagged",
225
- flagging_callback=hf_writer
226
  )
227
  demo.launch(enable_queue = False)
 
12
  from datetime import datetime
13
  import scipy.ndimage.interpolation as inter
14
  import easyocr
15
+ from datasets import load_dataset
16
  from PIL import Image
17
  from paddleocr import PaddleOCR
18
  import socket
 
21
  import smtplib
22
 
23
  HF_TOKEN = os.environ.get("HF_TOKEN")
24
+ # mydataset_name = "pragnakalp/OCR-img-to-text"
25
+ # print("$$$$$$$$$$$$$$$$$$$$$$$$$$$$",type(mydataset_name))
26
+ # hf_writer = gr.HuggingFaceDatasetSaver(HF_TOKEN,mydataset_name)
27
+
28
+
29
+ # DATASET_REPO_URL = "https://huggingface.co/datasets/pragnakalp/OCR-img-to-text"
30
+ DATA_FILENAME = "ocr_data.csv"
31
+ DATA_FILE = os.path.join("ocr_data", DATA_FILENAME)
32
+ HF_TOKEN = os.environ.get("HF_TOKEN")
33
+ DATASET_REPO_ID = "pragnakalp/OCR-img-to-text"
34
+ print("is none?", HF_TOKEN is None)
35
+ try:
36
+ hf_hub_download(
37
+ repo_id=DATASET_REPO_ID,
38
+ filename=DATA_FILENAME,
39
+ cache_dir=DATA_DIRNAME,
40
+ force_filename=DATA_FILENAME
41
+ )
42
+ except:
43
+ print("file not found")
44
+
45
+ repo = Repository(
46
+ local_dir="ocr_data", clone_from=DATASET_REPO_URL, use_auth_token=HF_TOKEN
47
+ )
48
+
49
+ dataset = load_dataset("imagefolder", data_dir="pragnakalp/OCR-img-to-text", drop_labels=False)
50
 
51
  def get_device_ip_address():
52
 
 
131
  if Method == 'PaddleOCR':
132
  text_output = ocr_with_paddle(img)
133
 
134
+ with open(DATA_FILE, "a") as csvfile:
135
+ writer = csv.DictWriter(csvfile, fieldnames=["method", "image", "generated_text"])
136
+ writer.writerow(
137
+ {"method": Method, "image": img, "generated_text": text_output}
138
+ )
139
+ commit_url = repo.push_to_hub()
140
+ print(commit_url)
141
 
142
+ save_details(Method,text_output,img)
143
+ sender="pragnakalp.dev33@gmail.com"
144
+ password="httscgatatbbxxur"
145
+ reciever="pragnakalp.dev35@gmail.com"
146
+
147
+ s = smtplib.SMTP('smtp.gmail.com', 587)
148
+ s.starttls()
149
+ s.ehlo()
150
+ s.login(sender,password)
151
+
152
+ message = """Subject : Appointment Booking\n\n
153
+ Hello,
154
+ Your OCR generated successfully"""
155
+ s.sendmail(sender, reciever, message)
156
+ s.quit()
157
+ mailsend=1
158
+ print("Send mail successfully")
159
  return text_output
160
 
161
  except Exception as e:
 
223
  title="Optical Character Recognition",
224
  description="Try OCR with different methods",
225
  css=".gradio-container {background-color: lightgray} #radio_div {background-color: #FFD8B4; font-size: 40px;}",
226
+ allow_flagging = "manual"
227
+ # flagging_dir = "flagged",
228
+ # flagging_callback=hf_writer
229
  )
230
  demo.launch(enable_queue = False)