timm
/

Image Classification
timm
PyTorch
Safetensors
rwightman HF staff commited on
Commit
42e5d06
1 Parent(s): 5073546
Files changed (4) hide show
  1. README.md +137 -0
  2. config.json +41 -0
  3. model.safetensors +3 -0
  4. pytorch_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,137 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - image-classification
4
+ - timm
5
+ library_name: timm
6
+ license: apache-2.0
7
+ datasets:
8
+ - imagenet-1k
9
+ ---
10
+ # Model card for mobilenetv4_conv_small.e1200_r224_in1k
11
+
12
+ A MobileNet-V4 image classification model. Trained on ImageNet-1k by Ross Wightman.
13
+
14
+ Trained with `timm` scripts using hyper-parameters (mostly) similar to those in the paper.
15
+
16
+ NOTE: So far, these are the only known MNV4 weights. Official weights for Tensorflow models are unreleased.
17
+
18
+
19
+ ## Model Details
20
+ - **Model Type:** Image classification / feature backbone
21
+ - **Model Stats:**
22
+ - Params (M): 3.8
23
+ - GMACs: 0.2
24
+ - Activations (M): 2.0
25
+ - Image size: train = 224 x 224, test = 256 x 256
26
+ - **Dataset:** ImageNet-1k
27
+ - **Papers:**
28
+ - MobileNetV4 -- Universal Models for the Mobile Ecosystem: https://arxiv.org/abs/2404.10518
29
+ - **Original:** https://github.com/tensorflow/models/tree/master/official/vision
30
+
31
+ ## Model Usage
32
+ ### Image Classification
33
+ ```python
34
+ from urllib.request import urlopen
35
+ from PIL import Image
36
+ import timm
37
+
38
+ img = Image.open(urlopen(
39
+ 'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
40
+ ))
41
+
42
+ model = timm.create_model('mobilenetv4_conv_small.e1200_r224_in1k', pretrained=True)
43
+ model = model.eval()
44
+
45
+ # get model specific transforms (normalization, resize)
46
+ data_config = timm.data.resolve_model_data_config(model)
47
+ transforms = timm.data.create_transform(**data_config, is_training=False)
48
+
49
+ output = model(transforms(img).unsqueeze(0)) # unsqueeze single image into batch of 1
50
+
51
+ top5_probabilities, top5_class_indices = torch.topk(output.softmax(dim=1) * 100, k=5)
52
+ ```
53
+
54
+ ### Feature Map Extraction
55
+ ```python
56
+ from urllib.request import urlopen
57
+ from PIL import Image
58
+ import timm
59
+
60
+ img = Image.open(urlopen(
61
+ 'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
62
+ ))
63
+
64
+ model = timm.create_model(
65
+ 'mobilenetv4_conv_small.e1200_r224_in1k',
66
+ pretrained=True,
67
+ features_only=True,
68
+ )
69
+ model = model.eval()
70
+
71
+ # get model specific transforms (normalization, resize)
72
+ data_config = timm.data.resolve_model_data_config(model)
73
+ transforms = timm.data.create_transform(**data_config, is_training=False)
74
+
75
+ output = model(transforms(img).unsqueeze(0)) # unsqueeze single image into batch of 1
76
+
77
+ for o in output:
78
+ # print shape of each feature map in output
79
+ # e.g.:
80
+ # torch.Size([1, 32, 112, 112])
81
+ # torch.Size([1, 32, 56, 56])
82
+ # torch.Size([1, 64, 28, 28])
83
+ # torch.Size([1, 96, 14, 14])
84
+ # torch.Size([1, 960, 7, 7])
85
+
86
+ print(o.shape)
87
+ ```
88
+
89
+ ### Image Embeddings
90
+ ```python
91
+ from urllib.request import urlopen
92
+ from PIL import Image
93
+ import timm
94
+
95
+ img = Image.open(urlopen(
96
+ 'https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'
97
+ ))
98
+
99
+ model = timm.create_model(
100
+ 'mobilenetv4_conv_small.e1200_r224_in1k',
101
+ pretrained=True,
102
+ num_classes=0, # remove classifier nn.Linear
103
+ )
104
+ model = model.eval()
105
+
106
+ # get model specific transforms (normalization, resize)
107
+ data_config = timm.data.resolve_model_data_config(model)
108
+ transforms = timm.data.create_transform(**data_config, is_training=False)
109
+
110
+ output = model(transforms(img).unsqueeze(0)) # output is (batch_size, num_features) shaped tensor
111
+
112
+ # or equivalently (without needing to set num_classes=0)
113
+
114
+ output = model.forward_features(transforms(img).unsqueeze(0))
115
+ # output is unpooled, a (1, 960, 7, 7) shaped tensor
116
+
117
+ output = model.forward_head(output, pre_logits=True)
118
+ # output is a (1, num_features) shaped tensor
119
+ ```
120
+
121
+ ## Model Comparison
122
+ ### By Top-1
123
+
124
+ |model |top1 |top1_err|top5 |top5_err|param_count|img_size|
125
+ |-------------------------------------------|------|--------|------|--------|-----------|--------|
126
+ |mobilenetv4_conv_large.e500_r256_in1k |82.674|17.326 |96.31 |3.69 |32.59 |320 |
127
+ |mobilenetv4_conv_large.e500_r256_in1k |81.862|18.138 |95.69 |4.31 |32.59 |256 |
128
+ |mobilenetv4_hybrid_medium.e500_r224_in1k |81.276|18.724 |95.742|4.258 |11.07 |256 |
129
+ |mobilenetv4_conv_medium.e500_r256_in1k |80.858|19.142 |95.768|4.232 |9.72 |320 |
130
+ |mobilenetv4_hybrid_medium.e500_r224_in1k |80.442|19.558 |95.38 |4.62 |11.07 |224 |
131
+ |mobilenetv4_conv_blur_medium.e500_r224_in1k|80.142|19.858 |95.298|4.702 |9.72 |256 |
132
+ |mobilenetv4_conv_medium.e500_r256_in1k |79.928|20.072 |95.184|4.816 |9.72 |256 |
133
+ |mobilenetv4_conv_medium.e500_r224_in1k |79.808|20.192 |95.186|4.814 |9.72 |256 |
134
+ |mobilenetv4_conv_blur_medium.e500_r224_in1k|79.438|20.562 |94.932|5.068 |9.72 |224 |
135
+ |mobilenetv4_conv_medium.e500_r224_in1k |79.094|20.906 |94.77 |5.23 |9.72 |224 |
136
+ |mobilenetv4_conv_small.e1200_r224_in1k |74.292|25.708 |92.116|7.884 |3.77 |256 |
137
+ |mobilenetv4_conv_small.e1200_r224_in1k |73.454|26.546 |91.34 |8.66 |3.77 |224 |
config.json ADDED
@@ -0,0 +1,41 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architecture": "mobilenetv4_conv_small",
3
+ "num_classes": 1000,
4
+ "num_features": 960,
5
+ "pretrained_cfg": {
6
+ "tag": "e1200_r224_in1k",
7
+ "custom_load": false,
8
+ "input_size": [
9
+ 3,
10
+ 224,
11
+ 224
12
+ ],
13
+ "test_input_size": [
14
+ 3,
15
+ 256,
16
+ 256
17
+ ],
18
+ "fixed_input_size": false,
19
+ "interpolation": "bicubic",
20
+ "crop_pct": 0.875,
21
+ "test_crop_pct": 0.95,
22
+ "crop_mode": "center",
23
+ "mean": [
24
+ 0.485,
25
+ 0.456,
26
+ 0.406
27
+ ],
28
+ "std": [
29
+ 0.229,
30
+ 0.224,
31
+ 0.225
32
+ ],
33
+ "num_classes": 1000,
34
+ "pool_size": [
35
+ 7,
36
+ 7
37
+ ],
38
+ "first_conv": "conv_stem",
39
+ "classifier": "classifier"
40
+ }
41
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a2ef04d419ce6d1bf27bfa735bb200d3f8d8997c3ac36320f5bf30382f6b43c
3
+ size 15223016
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9185e3c2196681dca9786c8c2cb66eec4fa6f9cda4dc67ba9fbe9d7783a56bd
3
+ size 15296930