Low-Cost Edge Computing Systems with New AI-based Hardware: Melanoma Detection Case

Mendoza-Tucto L.; Urbano K.; Auccahuasi W.; Castro-Mejia P.; Meza S.; Paucar-Miranda P. · Proceedings of the 9th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2025 (2025) · DOI: 10.1109/i-smac65734.2025.11597590

Resumen

The use of Artificial Intelligence is rapidly evolving, driven by the evolution in the use of different libraries for application design, primarily based on the Python language. Currently, technological solutions based on embedded hardware are being presented that integrate artificial intelligence processes such as training and detection, among others. This approach uses Huskylens hardware, which features an integrated computer vision system and classification, detection, and tracking algorithms, configured for melanoma detection using a skin lesion database. The procedures for training the device with its predefined functions are described. As a result, the detection of melanomas from the training database is presented, achieving a classification at 90% of the performance in the first one, as a recommendation, the procedures to perform performance tests with other databases are defined, as well as the possible uses that can be given in practical situations, where programming knowledge is not required to configure a solution based on artificial intelligence, in possible clinical uses.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad ESAN; Universidad Científica del Sur; Universidad César Vallejo; Universidad Autónoma de Ica; Universidad Privada del Norte