这篇教程Densely Connected Convolutional Network (DenseNet) keras实现代码写得很实用，希望能帮到您。
Densely Connected Convolutional Network (DenseNet) is a network architecture where each layer is directly connected to every other layer in a feed-forward fashion. It's quite similar to ResNet but in contrast DenseNet concatenates outputs instead of using summation. If you need a quick introduction about how DenseNet works, please read the original paper[1]. It's well written and easy to understand.
I implemented a DenseNet in Python using Keras and TensorFlow as backend. Because of this I can't guarantee that this implementation is working well with Theano or CNTK. In the next months I will update the code to TensorFlow 2.x. Besides I will try to optimize this architecture in my own way with some modifications. You can find several implementations on GitHub.
Results
Fashion-MNIST
I used this notebook to evaluate the model on fashion-MNIST with following parameters:
| Dense Blocks |
Depth |
Growth Rate |
Dropout |
Bottlen. |
Compress. |
BatchSize /
Epochs |
Training
(loss / acc) |
Validation
(loss / acc) |
Test
(loss / acc) |
| 5 |
35 |
20 |
0.4 |
False |
0.9 |
100 / 80 |
0.1366 / 0.9681 |
0.1675 / 0.9643 |
0.2739 / 0.9459 |
Feel free to try it on your own with another parameters.
Requirements
- TensorFlow 2.x
- Python 3.x
Usage
Feel free to use this implementation:
import densenet
model = densenet.DenseNet(input_shape=(28,28,1), nb_classes=10, depth=10, growth_rate=25,
dropout_rate=0.1, bottleneck=False, compression=0.5).build_model()
model.summary()
densenet一种keras改进的实现源代码
keras实现densenet源代码Keras Implementation of DenseNet |
