Skip to content

Commit a50003c

Browse files
james77777778james77777778
authored
Add Colab examples (#24)
* Update README * Update `README` * Update `README` * Update README * Update `README` * Update `README` * Update image * Update aligning
1 parent 2f771ac commit a50003c

File tree

1 file changed

+71
-43
lines changed

1 file changed

+71
-43
lines changed

‎README.md

Lines changed: 71 additions & 43 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,7 @@
44
# Keras Image Models
55

66
<div align="center">
7-
<img width="50%" src="docs/banner/kimm.png" alt="KIMM">
7+
<img width="50%" src="https://github.com/james77777778/kimm/assets/20734616/b21db8f2-307b-4791-b93d-e913e45fb238" alt="KIMM">
88

99
[![PyPI](https://img.shields.io/pypi/v/kimm)](https://pypi.org/project/kimm/)
1010
[![Contributions Welcome](https://img.shields.io/badge/contributions-welcome-brightgreen.svg?style=flat)](https://github.com/james77777778/kimm/issues)
@@ -24,10 +24,15 @@ pip install keras kimm
2424

2525
## Quickstart
2626

27-
### Use Pretrained Model
27+
### Image Classification Using the Model Pretrained on ImageNet
28+
29+
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/14WxYgVjlwCIO9MwqPYW-dskbTL2UHsVN?usp=sharing)
2830

2931
```python
30-
from keras import random
32+
import cv2
33+
import keras
34+
from keras import ops
35+
from keras.applications.imagenet_utils import decode_predictions
3136

3237
import kimm
3338

@@ -38,58 +43,81 @@ print(kimm.list_models())
3843
print(kimm.list_models("efficientnet", weights="imagenet")) # fuzzy search
3944

4045
# Initialize the model with pretrained weights
41-
model = kimm.models.EfficientNetV2B0(weights="imagenet")
42-
43-
# Predict
44-
x = random.uniform([1, 192, 192, 3]) * 255.0
45-
y = model.predict(x)
46-
print(y.shape)
46+
model = kimm.models.EfficientNetV2B0()
47+
image_size = model._default_size
4748

48-
# Initialize the model as a feature extractor with pretrained weights
49-
model = kimm.models.EfficientNetV2B0(
50-
feature_extractor=True, weights="imagenet"
49+
# Load an image as the model input
50+
image_path = keras.utils.get_file(
51+
"african_elephant.jpg", "https://i.imgur.com/Bvro0YD.png"
5152
)
53+
image = cv2.imread(image_path)
54+
image = cv2.resize(image, (image_size, image_size))
55+
x = ops.convert_to_tensor(image)
56+
x = ops.expand_dims(x, axis=0)
5257

53-
# Extract features for downstream tasks
54-
y = model.predict(x)
55-
print(y.keys())
56-
print(y["BLOCK5_S32"].shape)
58+
# Predict
59+
preds = model.predict(x)
60+
print("Predicted:", decode_predictions(preds, top=3)[0])
5761
```
5862

59-
### Transfer Learning
63+
```bash
64+
['ConvMixer1024D20', 'ConvMixer1536D20', 'ConvMixer736D32', 'ConvNeXtAtto', ...]
65+
['EfficientNetB0', 'EfficientNetB1', 'EfficientNetB2', 'EfficientNetB3', ...]
66+
1/1 ━━━━━━━━━━━━━━━━━━━━ 11s 11s/step
67+
Predicted: [('n02504458', 'African_elephant', 0.90578836), ('n01871265', 'tusker', 0.024864597), ('n02504013', 'Indian_elephant', 0.01161992)]
68+
```
6069

61-
```python
62-
from keras import layers
63-
from keras import models
64-
from keras import random
70+
### An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset
6571

66-
import kimm
72+
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1IbqfqG2NKEOKvBOznIPT1kjOdVPfThmd?usp=sharing)
6773

68-
# Initialize the model as a backbone with pretrained weights
69-
backbone = kimm.models.EfficientNetV2B0(
70-
input_shape=[224, 224, 3],
71-
include_top=False,
72-
pooling="avg",
73-
weights="imagenet",
74-
)
74+
Using `kimm.models.EfficientNetLiteB0`:
7575

76-
# Freeze the backbone for transfer learning
77-
backbone.trainable = False
76+
<div align="center">
77+
<img width="75%" src="https://github.com/james77777778/kimm/assets/20734616/cbfc0773-a3fa-407d-be9a-fba4f19da6d3" alt="kimm_prediction_0">
7878

79-
# Construct the model with new head
80-
inputs = layers.Input([224, 224, 3])
81-
x = backbone(inputs, training=False)
82-
x = layers.Dropout(0.2)(x)
83-
outputs = layers.Dense(2)(x)
84-
model = models.Model(inputs, outputs)
79+
<img width="75%" src="https://github.com/james77777778/kimm/assets/20734616/2eac0831-75bb-4790-a3af-412c3e09cf8f" alt="kimm_prediction_1">
80+
</div>
8581

86-
# Train the new model (put your own logic here)
82+
Reference: [Transfer learning & fine-tuning (keras.io)](https://keras.io/guides/transfer_learning/#an-endtoend-example-finetuning-an-image-classification-model-on-a-cats-vs-dogs-dataset)
8783

88-
# Predict
89-
x = random.uniform([1, 224, 224, 3]) * 255.0
90-
y = model.predict(x)
91-
print(y.shape)
92-
```
84+
### Grad-CAM
85+
86+
[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1h25VmsYDOLL6BNbRPEVOh1arIgcEoHu6?usp=sharing)
87+
88+
Using `kimm.models.MobileViTS`:
89+
90+
<div align="center">
91+
<img width="75%" src="https://github.com/james77777778/kimm/assets/20734616/cb5022a3-aaea-4324-a9cd-3d2e63a0a6b2" alt="grad_cam">
92+
</div>
93+
94+
Reference: [Grad-CAM class activation visualization (keras.io)](https://keras.io/examples/vision/grad_cam/)
95+
96+
## Model Zoo
97+
98+
|Model|Paper|Weights are ported from|API|
99+
|-|-|-|-|
100+
|ConvMixer|[ICLR 2022 Submission](https://arxiv.org/abs/2201.09792)|`timm`|`kimm.models.ConvMixer*`|
101+
|ConvNeXt|[CVPR 2022](https://arxiv.org/abs/2201.03545)|`timm`|`kimm.models.ConvNeXt*`|
102+
|DenseNet|[CVPR 2017](https://arxiv.org/abs/1608.06993)|`timm`|`kimm.models.DenseNet*`|
103+
|EfficientNet|[ICML 2019](https://arxiv.org/abs/1905.11946)|`timm`|`kimm.models.EfficientNet*`|
104+
|EfficientNetLite|[ICML 2019](https://arxiv.org/abs/1905.11946)|`timm`|`kimm.models.EfficientNetLite*`|
105+
|EfficientNetV2|[ICML 2021](https://arxiv.org/abs/2104.00298)|`timm`|`kimm.models.EfficientNetV2*`|
106+
|GhostNet|[CVPR 2020](https://arxiv.org/abs/1911.11907)|`timm`|`kimm.models.GhostNet*`|
107+
|GhostNetV2|[NeurIPS 2022](https://arxiv.org/abs/2211.12905)|`timm`|`kimm.models.GhostNetV2*`|
108+
|InceptionV3|[CVPR 2016](https://arxiv.org/abs/1512.00567)|`timm`|`kimm.models.InceptionV3`|
109+
|LCNet|[arXiv 2021](https://arxiv.org/abs/2109.15099)|`timm`|`kimm.models.LCNet*`|
110+
|MobileNetV2|[CVPR 2018](https://arxiv.org/abs/1801.04381)|`timm`|`kimm.models.MobileNetV2*`|
111+
|MobileNetV3|[ICCV 2019](https://arxiv.org/abs/1905.02244)|`timm`|`kimm.models.MobileNetV3*`|
112+
|MobileViT|[ICLR 2022](https://arxiv.org/abs/2110.02178)|`timm`|`kimm.models.MobileViT*`|
113+
|RegNet|[CVPR 2020](https://arxiv.org/abs/2003.13678)|`timm`|`kimm.models.RegNet*`|
114+
|ResNet|[CVPR 2015](https://arxiv.org/abs/1512.03385)|`timm`|`kimm.models.ResNet*`|
115+
|TinyNet|[NeurIPS 2020](https://arxiv.org/abs/2010.14819)|`timm`|`kimm.models.TinyNet*`|
116+
|VGG|[ICLR 2015](https://arxiv.org/abs/1409.1556)|`timm`|`kimm.models.VGG*`|
117+
|ViT|[ICLR 2021](https://arxiv.org/abs/2010.11929)|`timm`|`kimm.models.VisionTransformer*`|
118+
|Xception|[CVPR 2017](https://arxiv.org/abs/1610.02357)|`keras`|`kimm.models.Xception`|
119+
120+
The export scripts can be found in `tools/convert_*.py`.
93121

94122
## License
95123

0 commit comments

Comments
 (0)