DermaDetect seeks to leverage the latest in deep learning models paired with computer vision to create a compact, affordable, and accessible diagnostic tool for skin condition analysis, built on medically annotated data and with full transparency in mind. With the power of transfer learning techniques integrated with affordable yet capable hardware components like Raspberry Pi and a custom-designed rotating rig, we sought to provide a system capable of detecting common dermatological conditions with a good amount of precision.

 

DermaDetect employs cutting-edge and open-source methodologies, from its meticulous preprocessing of its public and professionally annotated datasets to its hardware design crafted to permit our capturing high-resolution facial images from multiple angles and maximizing the reliability of its results. Despite great and occasionally unexpected resource constraints, our team demonstrated a penchant for resilience and innovation, achieving significant milestones such as the successful training and deployment of our model on cost-effective equipment and delivering a polished product under great duress. While challenges such as real-world data variability and hardware tolerances remain, DermaDetect serves as a bedrock for deep learning prospects within clinical practice. This initiative underscores our commitment to employing engineering excellence in the service of impactful real-world applications and using technological development to shatter barriers of access to fundamental human rights such as medical care, in a small but incremental step in building a world meant for everyone.