The main aim of this project is to build a machine learning model capable of classifying systemic skin diseases from dermatological images using CNNs. By applying advanced architectures such as Inception, ResNet, DenseNet, or VGG through transfer learning, the system will extract image features and identify patterns associated with various skin conditions. The ultimate goal is to support medical professionals in diagnosing diseases more accurately and earlier, which can lead to better treatment outcomes. The project emphasizes learning deep learning workflows while contributing to real-world applications in healthcare, with the understanding that the model is assistive and cannot replace medical expertise.
This project follows a structured implementation plan. Students will begin by installing Python, setting up the development environment using tools like Anaconda or Google Colab, and learning the foundational structure of CNN models. The next steps involve gathering dermatology image datasets, preparing the data using augmentation and normalization techniques, and designing a basic model framework.
In the following weeks, students will train the model on the labeled data, test it with new images, evaluate its accuracy, and optimize performance using methods like dropout, batch normalization, and tuning hyperparameters. Toward the final weeks, students will complete the system’s development, conduct rigorous testing, document their work, and deliver a team presentation. Throughout the project, students must adhere to uniform coding ethics, maintain clear documentation, and avoid any form of plagiarism.
