Project Overview
Developed a multimodal deep learning solution for skin disease detection using both medical images and clinical text.
What I Did
* Prepared and preprocessed skin disease image data.
* Applied image preprocessing techniques suitable for deep learning models.
* Explored multiple CNN architectures for image feature extraction.
* Implemented ResNet50, MobileNet, and EfficientNet for image-based learning.
* Processed clinical text using transformer-based NLP models.
* Used BERT and RoBERTa to extract meaningful features from clinical descriptions.
* Combined image and text representations through a custom multimodal fusion approach.
* Built a model capable of learning from both visual and textual information.
Technologies
Python · TensorFlow · Deep Learning · Computer Vision · NLP · ResNet50 · MobileNet · EfficientNet · BERT · RoBERTa
Outcome
The project demonstrates a multimodal approach that combines image-based and text-based information to improve the learning process for skin disease classification.
