https://github.com/ashishpatel26/skin-lesions-detection-deep-learning
Transfer Learning with DCNNs (DenseNet, Inception V3, Inception-ResNet V2, VGG16) for skin lesions classification on HAM10000 dataset largescale data.
https://github.com/ashishpatel26/skin-lesions-detection-deep-learning
skin skin-cancer skin-cancer-detection skin-detection skin-disease skin-disease-classifiction skin-lesion-classification skin-lesion-segmentation skin-segmentation
Last synced: 5 months ago
JSON representation
Transfer Learning with DCNNs (DenseNet, Inception V3, Inception-ResNet V2, VGG16) for skin lesions classification on HAM10000 dataset largescale data.
- Host: GitHub
- URL: https://github.com/ashishpatel26/skin-lesions-detection-deep-learning
- Owner: ashishpatel26
- License: mit
- Created: 2020-12-01T13:12:30.000Z (almost 5 years ago)
- Default Branch: main
- Last Pushed: 2020-12-01T13:48:18.000Z (almost 5 years ago)
- Last Synced: 2025-04-02T17:52:56.370Z (6 months ago)
- Topics: skin, skin-cancer, skin-cancer-detection, skin-detection, skin-disease, skin-disease-classifiction, skin-lesion-classification, skin-lesion-segmentation, skin-segmentation
- Language: Jupyter Notebook
- Homepage:
- Size: 11.3 MB
- Stars: 70
- Watchers: 3
- Forks: 28
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Skin Lesions Classification with Deep Convolutional Neural Network
### **Dataset**
---
[**The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions**](https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T)
---
## Files Tables
| Sr.No | File Name | Link |
| ----- | ------------------------------------------------------------ | ------------------------------------------------------------ |
| 1 | Exploratory data analysis | [**Notebook**](http://nbviewer.ipython.org/urls/raw.github.com/ashishpatel26/Skin-Lesions-Detection-Deep-learning/main/Notebooks/Skin_Cancer_EDA.ipynb) |
| 2 | Baseline model | [**Notebook**](http://nbviewer.ipython.org/urls/raw.github.com/ashishpatel26/Skin-Lesions-Detection-Deep-learning/main/Notebooks/Baseline_CNN.ipynb) |
| 3 | Fine-tuning the last convolutional block of VGG16 | [**Notebook**](http://nbviewer.ipython.org/urls/raw.github.com/ashishpatel26/Skin-Lesions-Detection-Deep-learning/main/Notebooks/Fine_Tuning_VGG16.ipynb) |
| 4 | Fine-tuning the top 2 inception blocks of InceptionV3 | [**Notebook**](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Fine_Tuning_InceptionV3.ipynb) |
| 5 | Fine-tuning the Inception-ResNet-C of Inception-ResNet V2 | [**Notebook**](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Fine_Tuning_InceptionResNet.ipynb) |
| 6 | Fine-tuning the last dense block of DenseNet 201 | **[Notebook](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Fine_Tuning_DenseNet.ipynb)** |
| 7 | Fine-tuning all layers of pretrained Inception V3 on ImageNet | [**Notebook**](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Retraining_InceptionV3.ipynb) |
| 8 | Fine-tuning all layers of pretrained DenseNet 201 on ImageNet | [**Notebook**](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Retraining_DenseNet.ipynb) |
| 9 | Ensemble model of the fully fine-tuned Inception V3 and DenseNet 201 | **[Notebook](https://nbviewer.jupyter.org/github/ashishpatel26/Skin-Lesions-Detection-Deep-learning/blob/main/Notebooks/Ensemble_Models.ipynb)** |
## Technical Issue
All Notebooks using Keras 2.2.4 and Tensorflow 1.11. Batch-Norm layer in this version of Keras is implemented in a way that: during training your network will always use the mini-batch statistics either the BN layer is frozen or not; also during inference you will use the previously learned statistics of the frozen BN layers. As a result, if you fine-tune the top layers, their weights will be adjusted to the mean/variance of the new dataset. Nevertheless, during inference they will receive data which are scaled differently because the mean/variance of the original dataset will be used. Consequently, if use Keras's example codes for fine-tuning Inception V3 or any network with batch norm layer, the results will be very bad. Please refer to issue [#9965](https://github.com/keras-team/keras/pull/9965) and [#9214](https://github.com/keras-team/keras/issues/9214). One temporary solution is:
```python
for layer in pre_trained_model.layers:
if hasattr(layer, 'moving_mean') and hasattr(layer, 'moving_variance'):
layer.trainable = True
K.eval(K.update(layer.moving_mean, K.zeros_like(layer.moving_mean)))
K.eval(K.update(layer.moving_variance, K.zeros_like(layer.moving_variance)))
else:
layer.trainable = False
```
## Results
| Models | Validation | Test | Depth | # Params |
| ------------- |:-------------:| :-------------:| :-------------:| :-------------:|
| Baseline | 77.48% |76.54% | 11 layers | 2,124,839 |
| Fine-tuned VGG16 (from last block) | 79.82% | 79.64% | 23 layers | 14,980,935 |
| Fine-tuned Inception V3 (from the last 2 inception blocks) | 79.935% | 79.94% | 315 layers | 22,855,463 |
| Fine-tuned Inception-ResNet V2 (from the Inception-ResNet-C) | 80.82% | 82.53% | 784 layers | 55,127,271 |
| Fine-tuned DenseNet 201 (from the last dense block) | **85.8%** | **83.9%** | 711 layers | 19,309,127 |
| Fine-tuned Inception V3 (all layers) | 86.92% | 86.826% | _ | _ |
| Fine-tuned DenseNet 201 (all layers) | **86.696%** | **87.725%** | _ | _ |
| Ensemble of fully-fine-tuned Inception V3 and DenseNet 201 | **88.8%** | **88.52%** | _ | _ |
## The Dataset
[The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions](https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T,)