https://github.com/mozturan/imageclassification_hybridmodels

IMAGE CLASSIFICATION w/ hybrid classification algorithms
https://github.com/mozturan/imageclassification_hybridmodels

adaboost colab-notebook computer-vision deep-learning image-classification ipynb-jupyter-notebook knn machine-learning neural-networks pca satelite-images svm

Last synced: about 2 months ago
JSON representation

IMAGE CLASSIFICATION w/ hybrid classification algorithms

• Host: GitHub
• URL: https://github.com/mozturan/imageclassification_hybridmodels
• Owner: mozturan
• Created: 2023-03-15T17:52:29.000Z (almost 2 years ago)
• Default Branch: main
• Last Pushed: 2024-02-15T19:42:01.000Z (11 months ago)
• Last Synced: 2024-11-13T08:41:25.130Z (about 2 months ago)
• Topics: adaboost, colab-notebook, computer-vision, deep-learning, image-classification, ipynb-jupyter-notebook, knn, machine-learning, neural-networks, pca, satelite-images, svm
• Language: Jupyter Notebook
• Homepage:
• Size: 152 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# IMAGE CLASSIFICATION w/ hybrid classification algorithms

**Satellite image classification on different ml algorithms.**

----------------

RSI-CB256 dataset, which contains sensor images of four different classes, was used to test different hybrid classification algorithms. Since the images have two different resolutions, 224x224 and 64x64, all images were resized and converted to 224x224.

![Screenshot from 2024-02-15 22-27-53](https://github.com/9Xxi8Q4f/ML_ImageClassification/assets/89272933/4f321709-e9cb-4be1-a96d-1eee499afa31)

The architecture of the hybrid models was created as "CNN + classical ML models" and the CNN architecture was provided from the VGG16 architecture. A backbone created using transfer learning technique with CNN (VGG16) and ImageNet was then given to the fully connected layer and classification was made with neural networks. The outputs of the CNN layers were drawn and given to KNN, Logistic Regression, SVM, RF, AdaBoost classifiers, and the performance comparison of the classification of CNN outputs was made.

![Screenshot from 2024-02-15 22-32-01](https://github.com/9Xxi8Q4f/ML_ImageClassification/assets/89272933/6c0b53dd-a9de-4f3e-8a83-888cfa0dc8dc)

## **Training and validation loss and accuracy rate of deep network.**

![Screenshot from 2024-02-15 22-40-02](https://github.com/9Xxi8Q4f/ML_ImageClassification/assets/89272933/b6eba867-3fcd-4707-8a4e-9cc9d21eb518)

![Screenshot from 2024-02-15 22-41-44](https://github.com/9Xxi8Q4f/ML_ImageClassification/assets/89272933/7b9a2ae7-8e88-4f8a-88d5-0a412270d72f)