DermaSense helps clinicians and research teams standardize image intake, run consistent CNN inference, and document precautions and care guidance — with full audit trails per user account.
Images are resized, normalized, and scored through your deployed Keras model with calibrated confidence metrics.
Precautions and formulary hints are mapped per disease class and maintained centrally by your administrators.
Session-based access control ensures each user only sees their own uploads, history, and downloadable reports.
From capture to archived PDF-ready report in four controlled steps — ideal for capstone demonstrations and pilot clinics.
JPEG/PNG uploads with size and integrity checks; preview before submission reduces bad captures.
Images are standardized to model input tensors with consistent scaling for reproducible scores.
Each prediction stores disease label, softmax confidence, timestamp, and links to education content.
Print-optimized HTML reports support browser PDF export for charts and EMR attachment workflows.
This platform is intended for educational and decision-support contexts. Always validate model performance on your target population and follow local regulatory requirements for medical devices. Administrators can rotate model weights, remap class order, and purge obsolete records without touching unrelated tenants.
Train with curated public corpora such as the Skin Diseases image dataset on Kaggle, then upload your .keras/.h5 artifact and label map.
Provision a workspace