{"id":2201,"date":"2023-06-10T10:07:00","date_gmt":"2023-06-10T10:07:00","guid":{"rendered":"https:\/\/open.ieec.uned.es\/iot\/?p=2201"},"modified":"2023-07-27T10:12:35","modified_gmt":"2023-07-27T10:12:35","slug":"human-stability-assessment-and-fall-detection-based-on-dynamic-descriptors","status":"publish","type":"post","link":"https:\/\/open.ieec.uned.es\/iot\/human-stability-assessment-and-fall-detection-based-on-dynamic-descriptors\/","title":{"rendered":"Human stability assessment and fall detection based on dynamic descriptors"},"content":{"rendered":"\n

Recientemente han aprobado para publicaci\u00f3n en la revista \u00abIET Image Processing<\/em>\u00ab, indexada en JCR, el art\u00edculo titulado \u00abHuman stability assessment and fall detection based on dynamic descriptors<\/strong>\u00ab.<\/p>\n\n\n\n

Este trabajo es fruto de una tesis doctoral realizada dentro del programa de Doctorado en Tecnolog\u00edas Industriales, y dirigida por Sergio Mart\u00edn (Profesor Titular de la UNED) y V\u00edctor Rodr\u00edguez Ontiveros (investigador colaborador de la Universidad de Zaragoza).<\/p>\n\n\n\n

Aqu\u00ed os dejamos el abstract por si os resulta de inter\u00e9s:<\/p>\n\n\n\n

Both fall detection and fall prevention systems use an array of different technologies to<\/em> achieve their goals, contributing this way to better life conditions for the elderly<\/em> community. The artificial vision is one of these technologies and, within this field, it has<\/em> gained momentum over the course of the last few years with the incorporation of<\/em> different neural network architectures. These architectures, although different, share a<\/em> common characteristic, they are used to extract static or kinematic descriptors from<\/em> images and video clips that, properly processed, will determine if a fall has taken place<\/em> or if a patient\u2019s gait fulfils the characteristics associated to frequent fallers\u2019 ones.<\/em> However, these descriptors are inferred from datasets recorded by young volunteers or<\/em> actors who simulate falls, and the differences between these falls and the real ones are<\/em> well documented. This way, concerns about system performances in the wild, out of<\/em> laboratory environments, are raised.<\/em><\/p>\n\n\n\n

This work aims to reframe the problem of fall detection and approach it as a stability<\/em> assessment task. To do it, for the first time to the best of our knowledge, we propose<\/em> the introduction of human dynamic stability descriptors used in other fields. These new<\/em> descriptors are determined by using the information provided by a neural network able<\/em> to estimate the body center of mass and the feet projections onto the ground plane, as<\/em> well as the feet contact status with the ground.<\/em><\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The theory behind this new approach and its validity is studied in this article with very<\/em> promising results, as it is able to match or over exceed the performances of previous<\/em> systems using kinematic or static descriptors in laboratory conditions and, given the<\/em> independence of this system performances from fall dataset information, it has the<\/em> potential to behave better than systems based on other approaches in real world<\/em> environments.<\/em><\/p>\n\n\n\n

Y aqu\u00ed os dejamos la referencia por si quer\u00e9is citarla:<\/p>\n\n\n\n