[Issue #442] Facial Skin Diseases Classification using DL

[Radhika-789]

Issue Title

Facial Skin Diseases Classification using DL

• Info about the related issue (Aim of the project):
  Develop a deep learning-based skin disease classification system and compare multiple deep learning architectures on a multi-class dermatology dataset.

• Name: Radhika

Closes: #442

Describe the add-ons or changes you've made 📃

• Created a balanced 7-class subset from the DermNet dataset for skin disease classification.

• Dataset used: https://www.kaggle.com/datasets/shubhamgoel27/dermnet

• The original dataset provided in the issue contained only one class (Acne), which was not suitable for multi-class classification and model comparison. After discussion with the maintainer, a multi-class dataset was used.

• Implemented and trained seven deep learning models:

  • Custom CNN
  • VGG16
  • ResNet50
  • EfficientNetB0
  • MobileNetV2
  • DenseNet121
  • Xception

• Applied transfer learning using ImageNet pretrained weights and fine-tuned the last 30 layers of the pretrained architectures.

• Generated model comparison results using test accuracy and test loss.

• Added confusion matrices, classification reports, training history plots, and model comparison visualizations.

• Added project documentation, dataset information, requirements file, and repository structure.

Results

• Best Performing Model: VGG16
• Test Accuracy: 56.19%

Type of change ☑️

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

How Has This Been Tested? ⚙️

• Trained and evaluated all seven models on the prepared DermNet subset using Google Colab (T4 GPU).
• Compared model performance using test accuracy and test loss.
• Verified generated confusion matrices, classification reports, and training history plots.
• Confirmed successful execution of the complete notebook workflow.

Checklist: ☑️

• My code follows the guidelines of this project.
• I have performed a self-review of my own code.
• I have commented my code, particularly wherever it was hard to understand.
• I have made corresponding changes to the documentation.
• My changes generate no new warnings.
• I have added things that prove my feature works.
• Any dependent changes have been merged and published in downstream modules.

[github-actions]

Our team will soon review your PR. Thanks @Radhika-789 :)