Update README.md

#23
by putul - opened
Files changed (1) hide show
  1. README.md +8 -84
README.md CHANGED
@@ -1,90 +1,14 @@
1
- ---
2
- tags:
3
- - image-to-text
4
- - image-captioning
5
- license: apache-2.0
6
- widget:
7
- - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/savanna.jpg
8
- example_title: Savanna
9
- - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/football-match.jpg
10
- example_title: Football Match
11
- - src: https://huggingface.co/datasets/mishig/sample_images/resolve/main/airport.jpg
12
- example_title: Airport
13
- ---
14
 
15
- # nlpconnect/vit-gpt2-image-captioning
 
16
 
17
- This is an image captioning model trained by @ydshieh in [flax ](https://github.com/huggingface/transformers/tree/main/examples/flax/image-captioning) this is pytorch version of [this](https://huggingface.co/ydshieh/vit-gpt2-coco-en-ckpts).
18
 
 
19
 
20
- # The Illustrated Image Captioning using transformers
21
 
22
- ![](https://ankur3107.github.io/assets/images/vision-encoder-decoder.png)
23
 
24
- * https://ankur3107.github.io/blogs/the-illustrated-image-captioning-using-transformers/
25
-
26
-
27
- # Sample running code
28
-
29
- ```python
30
-
31
- from transformers import VisionEncoderDecoderModel, ViTImageProcessor, AutoTokenizer
32
- import torch
33
- from PIL import Image
34
-
35
- model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
36
- feature_extractor = ViTImageProcessor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
37
- tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
38
-
39
- device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
40
- model.to(device)
41
-
42
-
43
-
44
- max_length = 16
45
- num_beams = 4
46
- gen_kwargs = {"max_length": max_length, "num_beams": num_beams}
47
- def predict_step(image_paths):
48
- images = []
49
- for image_path in image_paths:
50
- i_image = Image.open(image_path)
51
- if i_image.mode != "RGB":
52
- i_image = i_image.convert(mode="RGB")
53
-
54
- images.append(i_image)
55
-
56
- pixel_values = feature_extractor(images=images, return_tensors="pt").pixel_values
57
- pixel_values = pixel_values.to(device)
58
-
59
- output_ids = model.generate(pixel_values, **gen_kwargs)
60
-
61
- preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
62
- preds = [pred.strip() for pred in preds]
63
- return preds
64
-
65
-
66
- predict_step(['doctor.e16ba4e4.jpg']) # ['a woman in a hospital bed with a woman in a hospital bed']
67
-
68
- ```
69
-
70
- # Sample running code using transformers pipeline
71
-
72
- ```python
73
-
74
- from transformers import pipeline
75
-
76
- image_to_text = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")
77
-
78
- image_to_text("https://ankur3107.github.io/assets/images/image-captioning-example.png")
79
-
80
- # [{'generated_text': 'a soccer game with a player jumping to catch the ball '}]
81
-
82
-
83
- ```
84
-
85
-
86
- # Contact for any help
87
- * https://huggingface.co/ankur310794
88
- * https://twitter.com/ankur310794
89
- * http://github.com/ankur3107
90
- * https://www.linkedin.com/in/ankur310794
 
1
+ What value will be printed?
 
 
 
 
 
 
 
 
 
 
 
 
2
 
3
+ code snippet image
4
+ (Assume necessary headers are included)
5
 
6
+ This code will always print 3, because the thread t is started before the value of y is modified.
7
 
8
+ This code won’t compile.
9
 
10
+ This code will always print 7, because the thread t is joined after both modifications of y.
11
 
12
+ This code will print 3, 6 or 7 (depending on thread scheduling by the operating system)
13
 
14
+ This code has undefined behavior.