• Implementation
• Backend: Flask app with REST endpoints for prediction, image upload, and result return.
• Frontend: HTML/CSS/Bootstrap interface for a clean user experience.
• Machine Learning:
o Model trained using Convolutional Neural Networks (CNN).
o Dataset: HAM10000 + custom-labeled dataset.
o Techniques: Data augmentation, transfer learning.
Screenshots: Homepage, image upload, result display, admin dashboard.
• Test Cases:
o Upload image → Should accept PNG, JPG, etc.
o Invalid input → Show error gracefully.
o Server response time → Less than 3 seconds.
• Testing Methods:
o Unit Testing (for model input/output).
o Integration Testing (UI to model).
o Regression Testing (after model retrain).
• Model Accuracy: Evaluated using confusion matrix, F1-score, and validation set accuracy (e.g., 89.7%).
• User Testing: Survey results from test users on accuracy and usability.
.
o Dataset: HAM10000 + custom-labeled dataset.
o Techniques: Data augmentation, transfer learning.
Screenshots: Homepage, image upload, result display, admin dashboard.
• Test Cases:
o Upload image → Should accept PNG, JPG, etc.
o Invalid input → Show error gracefully.
o Server response time → Less than 3 seconds.
• Testing Methods:
o Unit Testing (for model input/output).
o Integration Testing (UI to model).
o Regression Testing (after model retrain).
• Model Accuracy: Evaluated using confusion matrix, F1-score, and validation set accuracy (e.g., 89.7%).
• User Testing: Survey results from test users on accuracy and usability.
