README.md · convnext_tiny.in12k_ft_in1k

1

---

2

license: apache-2.0

3

library_name: timm

4

tags:

5

- image-classification

6

- timm

7

- transformers

8

datasets:

9

- imagenet-1k

10

- imagenet-12k

11

---

12

# Model card for convnext_tiny.in12k_ft_in1k

13

14

A ConvNeXt image classification model. Pretrained in `timm` on ImageNet-12k (a 11821 class subset of full ImageNet-22k) and fine-tuned on ImageNet-1k by Ross Wightman.

15

16

ImageNet-12k training done on TPUs thanks to support of the [TRC](https://sites.research.google/trc/about/) program.

17

18

Fine-tuning performed on 8x GPU [Lambda Labs](https://lambdalabs.com/) cloud instances.

19

20

21

## Model Details

22

- **Model Type:** Image classification / feature backbone

23

- **Model Stats:**

24

- Params (M): 28.6

25

- GMACs: 4.5

26

- Activations (M): 13.4

27

- Image size: train = 224 x 224, test = 288 x 288

28

- **Papers:**

29

- A ConvNet for the 2020s: https://arxiv.org/abs/2201.03545

30

- **Original:** https://github.com/huggingface/pytorch-image-models

31

- **Dataset:** ImageNet-1k

32

- **Pretrain Dataset:** ImageNet-12k

33

34

## Model Usage

35

### Image Classification

36

```python

37

from urllib.request import urlopen

38

from PIL import Image

39

import timm

40

41

img = Image.open(urlopen(

42

'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'

43

))

44

45

model = timm.create_model('convnext_tiny.in12k_ft_in1k', pretrained=True)

46

model = model.eval()

47

48

# get model specific transforms (normalization, resize)

49

data_config = timm.data.resolve_model_data_config(model)

50

transforms = timm.data.create_transform(**data_config, is_training=False)

51

52

output = model(transforms(img).unsqueeze(0)) # unsqueeze single image into batch of 1

53

54

top5_probabilities, top5_class_indices = torch.topk(output.softmax(dim=1) * 100, k=5)

55

```

56

57

### Feature Map Extraction

58

```python

59

from urllib.request import urlopen

60

from PIL import Image

61

import timm

62

63

img = Image.open(urlopen(

64

'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'

65

))

66

67

model = timm.create_model(

68

'convnext_tiny.in12k_ft_in1k',

69

pretrained=True,

70

features_only=True,

71

)

72

model = model.eval()

73

74

# get model specific transforms (normalization, resize)

75

data_config = timm.data.resolve_model_data_config(model)

76

transforms = timm.data.create_transform(**data_config, is_training=False)

77

78

output = model(transforms(img).unsqueeze(0)) # unsqueeze single image into batch of 1

79

80

for o in output:

81

# print shape of each feature map in output

82

# e.g.:

83

# torch.Size([1, 96, 56, 56])

84

# torch.Size([1, 192, 28, 28])

85

# torch.Size([1, 384, 14, 14])

86

# torch.Size([1, 768, 7, 7])

87

88

print(o.shape)

89

```

90

91

### Image Embeddings

92

```python

93

from urllib.request import urlopen

94

from PIL import Image

95

import timm

96

97

img = Image.open(urlopen(

98

'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'

99

))

100

101

model = timm.create_model(

102

'convnext_tiny.in12k_ft_in1k',

103

pretrained=True,

104

num_classes=0, # remove classifier nn.Linear

105

)

106

model = model.eval()

107

108

# get model specific transforms (normalization, resize)

109

data_config = timm.data.resolve_model_data_config(model)

110

transforms = timm.data.create_transform(**data_config, is_training=False)

111

112

output = model(transforms(img).unsqueeze(0)) # output is (batch_size, num_features) shaped tensor

113

114

# or equivalently (without needing to set num_classes=0)

115

116

output = model.forward_features(transforms(img).unsqueeze(0))

117

# output is unpooled, a (1, 768, 7, 7) shaped tensor

118

119

output = model.forward_head(output, pre_logits=True)

120

# output is a (1, num_features) shaped tensor

121

```

122

123

## Model Comparison

124

Explore the dataset and runtime metrics of this model in timm [model results](https://github.com/huggingface/pytorch-image-models/tree/main/results).

125

126

All timing numbers from eager model PyTorch 1.13 on RTX 3090 w/ AMP.

127

128

| model                                                                                                                        |top1  |top5  |img_size|param_count|gmacs |macts |samples_per_sec|batch_size|

129

|------------------------------------------------------------------------------------------------------------------------------|------|------|--------|-----------|------|------|---------------|----------|

130

| [convnextv2_huge.fcmae_ft_in22k_in1k_512](https://huggingface.co/timm/convnextv2_huge.fcmae_ft_in22k_in1k_512)               |88.848|98.742|512     |660.29     |600.81|413.07|28.58          |48        |

131

| [convnextv2_huge.fcmae_ft_in22k_in1k_384](https://huggingface.co/timm/convnextv2_huge.fcmae_ft_in22k_in1k_384)               |88.668|98.738|384     |660.29     |337.96|232.35|50.56          |64        |

132

| [convnext_xxlarge.clip_laion2b_soup_ft_in1k](https://huggingface.co/timm/convnext_xxlarge.clip_laion2b_soup_ft_in1k)         |88.612|98.704|256     |846.47     |198.09|124.45|122.45         |256       |

133

| [convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_384](https://huggingface.co/timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_384)             |88.312|98.578|384     |200.13     |101.11|126.74|196.84         |256       |

134

| [convnextv2_large.fcmae_ft_in22k_in1k_384](https://huggingface.co/timm/convnextv2_large.fcmae_ft_in22k_in1k_384)             |88.196|98.532|384     |197.96     |101.1 |126.74|128.94         |128       |

135

| [convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_320](https://huggingface.co/timm/convnext_large_mlp.clip_laion2b_soup_ft_in12k_in1k_320)             |87.968|98.47 |320     |200.13     |70.21 |88.02 |283.42         |256       |

136

| [convnext_xlarge.fb_in22k_ft_in1k_384](https://huggingface.co/timm/convnext_xlarge.fb_in22k_ft_in1k_384)                     |87.75 |98.556|384     |350.2      |179.2 |168.99|124.85         |192       |

137

| [convnextv2_base.fcmae_ft_in22k_in1k_384](https://huggingface.co/timm/convnextv2_base.fcmae_ft_in22k_in1k_384)               |87.646|98.422|384     |88.72      |45.21 |84.49 |209.51         |256       |

138

| [convnext_large.fb_in22k_ft_in1k_384](https://huggingface.co/timm/convnext_large.fb_in22k_ft_in1k_384)                       |87.476|98.382|384     |197.77     |101.1 |126.74|194.66         |256       |

139

| [convnext_large_mlp.clip_laion2b_augreg_ft_in1k](https://huggingface.co/timm/convnext_large_mlp.clip_laion2b_augreg_ft_in1k) |87.344|98.218|256     |200.13     |44.94 |56.33 |438.08         |256       |

140

| [convnextv2_large.fcmae_ft_in22k_in1k](https://huggingface.co/timm/convnextv2_large.fcmae_ft_in22k_in1k)                     |87.26 |98.248|224     |197.96     |34.4  |43.13 |376.84         |256       |

141

| [convnext_base.clip_laion2b_augreg_ft_in12k_in1k_384](https://huggingface.co/timm/convnext_base.clip_laion2b_augreg_ft_in12k_in1k_384)                   |87.138|98.212|384     |88.59      |45.21 |84.49 |365.47         |256       |

142

| [convnext_xlarge.fb_in22k_ft_in1k](https://huggingface.co/timm/convnext_xlarge.fb_in22k_ft_in1k)                             |87.002|98.208|224     |350.2      |60.98 |57.5  |368.01         |256       |

143

| [convnext_base.fb_in22k_ft_in1k_384](https://huggingface.co/timm/convnext_base.fb_in22k_ft_in1k_384)                         |86.796|98.264|384     |88.59      |45.21 |84.49 |366.54         |256       |

144

| [convnextv2_base.fcmae_ft_in22k_in1k](https://huggingface.co/timm/convnextv2_base.fcmae_ft_in22k_in1k)                       |86.74 |98.022|224     |88.72      |15.38 |28.75 |624.23         |256       |

145

| [convnext_large.fb_in22k_ft_in1k](https://huggingface.co/timm/convnext_large.fb_in22k_ft_in1k)                               |86.636|98.028|224     |197.77     |34.4  |43.13 |581.43         |256       |

146

| [convnext_base.clip_laiona_augreg_ft_in1k_384](https://huggingface.co/timm/convnext_base.clip_laiona_augreg_ft_in1k_384)     |86.504|97.97 |384     |88.59      |45.21 |84.49 |368.14         |256       |

147

| [convnext_base.clip_laion2b_augreg_ft_in12k_in1k](https://huggingface.co/timm/convnext_base.clip_laion2b_augreg_ft_in12k_in1k)                           |86.344|97.97 |256     |88.59      |20.09 |37.55 |816.14         |256       |

148

| [convnextv2_huge.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_huge.fcmae_ft_in1k)                                   |86.256|97.75 |224     |660.29     |115.0 |79.07 |154.72         |256       |

149

| [convnext_small.in12k_ft_in1k_384](https://huggingface.co/timm/convnext_small.in12k_ft_in1k_384)                             |86.182|97.92 |384     |50.22      |25.58 |63.37 |516.19         |256       |

150

| [convnext_base.clip_laion2b_augreg_ft_in1k](https://huggingface.co/timm/convnext_base.clip_laion2b_augreg_ft_in1k)           |86.154|97.68 |256     |88.59      |20.09 |37.55 |819.86         |256       |

151

| [convnext_base.fb_in22k_ft_in1k](https://huggingface.co/timm/convnext_base.fb_in22k_ft_in1k)                                 |85.822|97.866|224     |88.59      |15.38 |28.75 |1037.66        |256       |

152

| [convnext_small.fb_in22k_ft_in1k_384](https://huggingface.co/timm/convnext_small.fb_in22k_ft_in1k_384)                       |85.778|97.886|384     |50.22      |25.58 |63.37 |518.95         |256       |

153

| [convnextv2_large.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_large.fcmae_ft_in1k)                                 |85.742|97.584|224     |197.96     |34.4  |43.13 |375.23         |256       |

154

| [convnext_small.in12k_ft_in1k](https://huggingface.co/timm/convnext_small.in12k_ft_in1k)                                     |85.174|97.506|224     |50.22      |8.71  |21.56 |1474.31        |256       |

155

| [convnext_tiny.in12k_ft_in1k_384](https://huggingface.co/timm/convnext_tiny.in12k_ft_in1k_384)                               |85.118|97.608|384     |28.59      |13.14 |39.48 |856.76         |256       |

156

| [convnextv2_tiny.fcmae_ft_in22k_in1k_384](https://huggingface.co/timm/convnextv2_tiny.fcmae_ft_in22k_in1k_384)               |85.112|97.63 |384     |28.64      |13.14 |39.48 |491.32         |256       |

157

| [convnextv2_base.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_base.fcmae_ft_in1k)                                   |84.874|97.09 |224     |88.72      |15.38 |28.75 |625.33         |256       |

158

| [convnext_small.fb_in22k_ft_in1k](https://huggingface.co/timm/convnext_small.fb_in22k_ft_in1k)                               |84.562|97.394|224     |50.22      |8.71  |21.56 |1478.29        |256       |

159

| [convnext_large.fb_in1k](https://huggingface.co/timm/convnext_large.fb_in1k)                                                 |84.282|96.892|224     |197.77     |34.4  |43.13 |584.28         |256       |

160

| [convnext_tiny.in12k_ft_in1k](https://huggingface.co/timm/convnext_tiny.in12k_ft_in1k)                                       |84.186|97.124|224     |28.59      |4.47  |13.44 |2433.7         |256       |

161

| [convnext_tiny.fb_in22k_ft_in1k_384](https://huggingface.co/timm/convnext_tiny.fb_in22k_ft_in1k_384)                         |84.084|97.14 |384     |28.59      |13.14 |39.48 |862.95         |256       |

162

| [convnextv2_tiny.fcmae_ft_in22k_in1k](https://huggingface.co/timm/convnextv2_tiny.fcmae_ft_in22k_in1k)                       |83.894|96.964|224     |28.64      |4.47  |13.44 |1452.72        |256       |

163

| [convnext_base.fb_in1k](https://huggingface.co/timm/convnext_base.fb_in1k)                                                   |83.82 |96.746|224     |88.59      |15.38 |28.75 |1054.0         |256       |

164

| [convnextv2_nano.fcmae_ft_in22k_in1k_384](https://huggingface.co/timm/convnextv2_nano.fcmae_ft_in22k_in1k_384)               |83.37 |96.742|384     |15.62      |7.22  |24.61 |801.72         |256       |

165

| [convnext_small.fb_in1k](https://huggingface.co/timm/convnext_small.fb_in1k)                                                 |83.142|96.434|224     |50.22      |8.71  |21.56 |1464.0         |256       |

166

| [convnextv2_tiny.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_tiny.fcmae_ft_in1k)                                   |82.92 |96.284|224     |28.64      |4.47  |13.44 |1425.62        |256       |

167

| [convnext_tiny.fb_in22k_ft_in1k](https://huggingface.co/timm/convnext_tiny.fb_in22k_ft_in1k)                                 |82.898|96.616|224     |28.59      |4.47  |13.44 |2480.88        |256       |

168

| [convnext_nano.in12k_ft_in1k](https://huggingface.co/timm/convnext_nano.in12k_ft_in1k)                                       |82.282|96.344|224     |15.59      |2.46  |8.37  |3926.52        |256       |

169

| [convnext_tiny_hnf.a2h_in1k](https://huggingface.co/timm/convnext_tiny_hnf.a2h_in1k)                                         |82.216|95.852|224     |28.59      |4.47  |13.44 |2529.75        |256       |

170

| [convnext_tiny.fb_in1k](https://huggingface.co/timm/convnext_tiny.fb_in1k)                                                   |82.066|95.854|224     |28.59      |4.47  |13.44 |2346.26        |256       |

171

| [convnextv2_nano.fcmae_ft_in22k_in1k](https://huggingface.co/timm/convnextv2_nano.fcmae_ft_in22k_in1k)                       |82.03 |96.166|224     |15.62      |2.46  |8.37  |2300.18        |256       |

172

| [convnextv2_nano.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_nano.fcmae_ft_in1k)                                   |81.83 |95.738|224     |15.62      |2.46  |8.37  |2321.48        |256       |

173

| [convnext_nano_ols.d1h_in1k](https://huggingface.co/timm/convnext_nano_ols.d1h_in1k)                                         |80.866|95.246|224     |15.65      |2.65  |9.38  |3523.85        |256       |

174

| [convnext_nano.d1h_in1k](https://huggingface.co/timm/convnext_nano.d1h_in1k)                                                 |80.768|95.334|224     |15.59      |2.46  |8.37  |3915.58        |256       |

175

| [convnextv2_pico.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_pico.fcmae_ft_in1k)                                   |80.304|95.072|224     |9.07       |1.37  |6.1   |3274.57        |256       |

176

| [convnext_pico.d1_in1k](https://huggingface.co/timm/convnext_pico.d1_in1k)                                                   |79.526|94.558|224     |9.05       |1.37  |6.1   |5686.88        |256       |

177

| [convnext_pico_ols.d1_in1k](https://huggingface.co/timm/convnext_pico_ols.d1_in1k)                                           |79.522|94.692|224     |9.06       |1.43  |6.5   |5422.46        |256       |

178

| [convnextv2_femto.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_femto.fcmae_ft_in1k)                                 |78.488|93.98 |224     |5.23       |0.79  |4.57  |4264.2         |256       |

179

| [convnext_femto_ols.d1_in1k](https://huggingface.co/timm/convnext_femto_ols.d1_in1k)                                         |77.86 |93.83 |224     |5.23       |0.82  |4.87  |6910.6         |256       |

180

| [convnext_femto.d1_in1k](https://huggingface.co/timm/convnext_femto.d1_in1k)                                                 |77.454|93.68 |224     |5.22       |0.79  |4.57  |7189.92        |256       |

181

| [convnextv2_atto.fcmae_ft_in1k](https://huggingface.co/timm/convnextv2_atto.fcmae_ft_in1k)                                   |76.664|93.044|224     |3.71       |0.55  |3.81  |4728.91        |256       |

182

| [convnext_atto_ols.a2_in1k](https://huggingface.co/timm/convnext_atto_ols.a2_in1k)                                           |75.88 |92.846|224     |3.7        |0.58  |4.11  |7963.16        |256       |

183

| [convnext_atto.d2_in1k](https://huggingface.co/timm/convnext_atto.d2_in1k)                                                   |75.664|92.9  |224     |3.7        |0.55  |3.81  |8439.22        |256       |

184

185

## Citation

186

```bibtex

187

@misc{rw2019timm,

188

author = {Ross Wightman},

189

title = {PyTorch Image Models},

190

year = {2019},

191

publisher = {GitHub},

192

journal = {GitHub repository},

193

doi = {10.5281/zenodo.4414861},

194

howpublished = {\url{https://github.com/huggingface/pytorch-image-models}}

195

}

196

```

197

```bibtex

198

@article{liu2022convnet,

199

author = {Zhuang Liu and Hanzi Mao and Chao-Yuan Wu and Christoph Feichtenhofer and Trevor Darrell and Saining Xie},

200

title = {A ConvNet for the 2020s},

201

journal = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},

202

year = {2022},

203

}

204

```

205