liminghao1630 commited on
Commit
34306f7
1 Parent(s): 25fe3e3

Update code example

Browse files
Files changed (1) hide show
  1. README.md +4 -9
README.md CHANGED
@@ -23,7 +23,7 @@ You can use the raw model for optical character recognition (OCR) on single text
23
  Here is how to use this model in PyTorch:
24
 
25
  ```python
26
- from transformers import TrOCRProcessor, VisionEncoderDecoderModel, AutoFeatureExtractor, XLMRobertaTokenizer
27
  from PIL import Image
28
  import requests
29
 
@@ -31,17 +31,12 @@ import requests
31
  url = 'https://fki.tic.heia-fr.ch/static/img/a01-122-02-00.jpg'
32
  image = Image.open(requests.get(url, stream=True).raw).convert("RGB")
33
 
34
- # For the time being, TrOCRProcessor does not support the small models, so the following temporary solution can be adopted
35
- # processor = TrOCRProcessor.from_pretrained('microsoft/trocr-small-printed')
36
- feature_extractor = AutoFeatureExtractor.from_pretrained('microsoft/trocr-small-printed')
37
- tokenizer = XLMRobertaTokenizer.from_pretrained('microsoft/trocr-small-printed')
38
  model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-small-printed')
39
- # pixel_values = processor(images=image, return_tensors="pt").pixel_values
40
- pixel_values = feature_extractor(images=image, return_tensors="pt").pixel_values
41
 
42
  generated_ids = model.generate(pixel_values)
43
- # generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
44
- generated_text = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
45
  ```
46
 
47
  ### BibTeX entry and citation info
 
23
  Here is how to use this model in PyTorch:
24
 
25
  ```python
26
+ from transformers import TrOCRProcessor, VisionEncoderDecoderModel
27
  from PIL import Image
28
  import requests
29
 
 
31
  url = 'https://fki.tic.heia-fr.ch/static/img/a01-122-02-00.jpg'
32
  image = Image.open(requests.get(url, stream=True).raw).convert("RGB")
33
 
34
+ processor = TrOCRProcessor.from_pretrained('microsoft/trocr-small-printed')
 
 
 
35
  model = VisionEncoderDecoderModel.from_pretrained('microsoft/trocr-small-printed')
36
+ pixel_values = processor(images=image, return_tensors="pt").pixel_values
 
37
 
38
  generated_ids = model.generate(pixel_values)
39
+ generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
 
40
  ```
41
 
42
  ### BibTeX entry and citation info