An Edge–Cloud Collaborative Scalp Inspection System Based on Robust Representation Learning

In personal healthcare, scalp health is often overlooked despite its critical role in systemic health. Traditional manual diagnostic methods for scalp conditions are invasive and subjective, resulting in variability and discomfort. This paper introduces the first Internet of Things (IoT)-based intelligent scalp inspection system leveraging edge-cloud collaborative computing to provide autonomous and real-time scalp health analysis using videocapillaroscope images. The system incorporates a new data augmentation strategy, Curriculum Enhanced Localizable Swap and Rotation (CE-LoSwRot), which mitigates data imbalance issues, and robust representation learning through Modified Balanced Contrastive Learning (M-BCL) and Test-Time Adaptation (TTA) to enhance feature representation and model adaptability. By integrating edge and cloud computing, the proposed system reduces network latency and congestion while ensuring real-time data processing and analysis. The comprehensive scalp problem classification dataset collected for this study demonstrates the system’s superior performance compared to existing state-of-the-art methods, establishing a new benchmark in scalp health diagnostics.