nielsr HF staff commited on
Commit
61d034f
1 Parent(s): ef5ff94

First draft of model card

Browse files
Files changed (1) hide show
  1. README.md +89 -0
README.md ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ datasets:
4
+ - imagenet-21k
5
+ ---
6
+
7
+ # Vision Transformer base model
8
+
9
+ Vision Transformer (ViT) model pre-trained on ImageNet-21k (14 million images, 21,843 classes) at resolution 224x224. It was introduced in the paper [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://arxiv.org/abs/2010.11929) by Dosovitskiy et al. and first released in [this repository](https://github.com/google-research/vision_transformer). However, the weights were converted from the [timm repository](https://github.com/rwightman/pytorch-image-models) by Ross Wightman, who already converted the weights from JAX to PyTorch. Credits go to him.
10
+
11
+ Disclaimer: The team releasing ViT did not write a model card for this model so this model card has been written by the Hugging Face team.
12
+
13
+ ## Model description
14
+
15
+ The Vision Transformer (ViT) is a transformer encoder model (BERT-like) pretrained on a large collection of images in a supervised fashion, namely ImageNet-21k, at a resolution of 224x224 pixels.
16
+
17
+ Images are presented to the model as a sequence of fixed-size patches (resolution 16x16), which are linearly embedded. One also adds a [CLS] token to the beginning of a sequence to use it for classification tasks. One also adds absolute position embeddings before feeding the sequence to the layers of the Transformer encoder.
18
+
19
+ By pre-training the model, it learns an inner representation of images that can then be used to extract features useful for downstream tasks: if you have a dataset of labeled images for instance, you can train a standard classifier by placing a linear layer on top of the pre-trained encoder. One typically places a linear layer on top of the [CLS] token, as the last hidden state of this token can be seen as a representation of an entire image.
20
+
21
+ ## Intended uses & limitations
22
+
23
+ You can use the raw model for image classification. See the [model hub](https://huggingface.co/models?search=google/vit) to look for
24
+ fine-tuned versions on a task that interests you.
25
+
26
+ ### How to use
27
+
28
+ Here is how to use this model to classify an image of the COCO 2017 dataset into one of the 21k ImageNet-21k classes:
29
+
30
+ ```python
31
+ from transformers import ViTFeatureExtractor, ViTForImageClassification
32
+ from PIL import Image
33
+ import requests
34
+ url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
35
+ image = Image.open(requests.get(url, stream=True).raw)
36
+ feature_extractor = ViTFeatureExtractor.from_pretrained('google/vit-base-patch16-224-in21k')
37
+ model = ViTForImageClassification.from_pretrained('google/vit-base-patch16-224-in21k')
38
+ inputs = feature_extractor(images=image)
39
+ outputs = model(**inputs)
40
+ logits = outputs.logits
41
+ # model predicts one of the 21k ImageNet-21k classes
42
+ predicted_class = logits.argmax(-1).item()
43
+ ```
44
+
45
+ Currently, both the feature extractor and model support PyTorch. Tensorflow and JAX/FLAX are coming soon, and the API of ViTFeatureExtractor might change.
46
+
47
+ ## Training data
48
+
49
+ The ViT model was pretrained on [ImageNet-21k](http://www.image-net.org/), a dataset consisting of 14 million images and 21k classes.
50
+
51
+ ## Training procedure
52
+
53
+ ### Preprocessing
54
+
55
+ The exact details of preprocessing of images during training/validation can be found [here](https://github.com/google-research/vision_transformer/blob/master/vit_jax/input_pipeline.py).
56
+
57
+ Images are resized/rescaled to the same resolution (224x224) and normalized across the RGB channels with mean (0.5, 0.5, 0.5) and standard deviation (0.5, 0.5, 0.5).
58
+
59
+ ### Pretraining
60
+
61
+ The model was trained on TPUv3 hardware (8 cores). All model variants are trained with a batch size of 4096 and learning rate warmup of 10k steps. For ImageNet, the authors found it beneficial to additionally apply gradient clipping at global norm 1. Pre-training resolution is 224.
62
+
63
+ ## Evaluation results
64
+
65
+ For evaluation results on several image classification benchmarks, we refer to tables 2 and 5 of the original paper. Note that for fine-tuning, the best results are obtained with a higher resolution (384x384). Of course, increasing the model size will result in better performance.
66
+
67
+ ### BibTeX entry and citation info
68
+
69
+ ```bibtex
70
+ @misc{wu2020visual,
71
+ title={Visual Transformers: Token-based Image Representation and Processing for Computer Vision},
72
+ author={Bichen Wu and Chenfeng Xu and Xiaoliang Dai and Alvin Wan and Peizhao Zhang and Zhicheng Yan and Masayoshi Tomizuka and Joseph Gonzalez and Kurt Keutzer and Peter Vajda},
73
+ year={2020},
74
+ eprint={2006.03677},
75
+ archivePrefix={arXiv},
76
+ primaryClass={cs.CV}
77
+ }
78
+ ```
79
+
80
+ ```bibtex
81
+ @inproceedings{deng2009imagenet,
82
+ title={Imagenet: A large-scale hierarchical image database},
83
+ author={Deng, Jia and Dong, Wei and Socher, Richard and Li, Li-Jia and Li, Kai and Fei-Fei, Li},
84
+ booktitle={2009 IEEE conference on computer vision and pattern recognition},
85
+ pages={248--255},
86
+ year={2009},
87
+ organization={Ieee}
88
+ }
89
+ ```