AIProject

Project Description

CNN Classifier for dataset: https://www.kaggle.com/datasets/ismailpromus/skin-diseases-image-dataset/data

\Proj Folder | - \data | | - Class 0 | | - Class 1 | | - Class 2 | | - Class 3 | | - Class 4 | | - Class 5 | | - Class 6 | | - Class 7 | | - Class 8 | | - Class 9 | - dataProcessing.py | - README.MD | - test1.ipynb

Task: Supervised Classification task with a CNN, using Transfer Learning on a EfficientNet Model

Workflow:

1. DataCleaning/Processing
   1. DataSet Splitting -> 90 : 10
   2. For Training Set: 90%
      1. Possibility of removing Watermarks?
      2. Padding To Square(resizing),
      3. DownSizing images (to reduce memory requirements)
      4. Normalizing Channels (for faster Learning)
   3. For Test Set: 10%
      1. No Augmentation, only resizing(Padding to Square) to fit the model inputs. use it raw to test for accuracy and generalization
2. Model Building/Training
   1. Importing Model from torchvision
   2. Freezing of all layers except the classification layer.
   3. Train Model and Assess performance

      1. Epoch, LR, Loss Fn, Additional Metrics

      2. IF BAD: Modify/Reselect Model

      3. IF GOOD: Progressive Unfreezing of Params for deeper learning of features.

Setup Instructions

1. Clone the Repository

git clone https://github.com/wsnc0/AIProject.git
cd AIProject/gui

2. Install Dependencies

pip install -r requirements.txt

3. Run the Application

python -m streamlit run app.py

Disclaimer

First Run On first run, the application will automatically download the AI model (~185MB) from Google Drive. This may take a few minutes depending on your internet connection.

Using the Application

1. Home Screen:

Click the "Start" button to begin Toggle "Use Gemini AI for enhanced descriptions" if you want detailed condition information

2. Upload Screen:

Drag and drop or browse to select a skin condition image Click "Analyze Image" to process the upload

3. Results Screen:

View the predicted skin condition and confidence score Examine the GradCAM heatmap showing areas of interest Read the enhanced description of the condition (if Gemini AI is enabled) Click "Start Over" to analyze another image

Supported Skin Conditions

The classifier can identify 10 different skin conditions: Eczema, Warts Molluscum, Melanoma, Atopic Dermatitis, Basal Cell Carcinoma, Melanocytic Nevi, Benign Keratosis, Psoriasis, Seborrheic Keratoses, Tinea Ringworm