Modelo de visión por computador basado en Red Neuronal DenseNet para el apoyo en la realización de ejercicios de yoga antiestrés.
Academic stress is a prevalent issue among university students. To address it, strategies such as conversational therapies, meditation, and therapeutic yoga exercises are implemented. However, la Unidad de Bienestar Universitario of Universidad Nacional de Loja (UNL) lacks the means to allow student...
-д хадгалсан:
| Үндсэн зохиолч: | |
|---|---|
| Формат: | bachelorThesis |
| Хэл сонгох: | spa |
| Хэвлэсэн: |
2024
|
| Нөхцлүүд: | |
| Онлайн хандалт: | https://dspace.unl.edu.ec/jspui/handle/123456789/30000 |
| Шошгууд: |
Шошго нэмэх
Шошго байхгүй, Энэхүү баримтыг шошголох эхний хүн болох!
|
| Тойм: | Academic stress is a prevalent issue among university students. To address it, strategies such as conversational therapies, meditation, and therapeutic yoga exercises are implemented. However, la Unidad de Bienestar Universitario of Universidad Nacional de Loja (UNL) lacks the means to allow students to engage in yoga practices as an alternative for stress reduction. In this context, the objective of this Curricular Integration Work is to develop a computer vision model based on a DenseNet Neural Network to support anti-stress yoga exercises assigned by the clinical psychologist of the Unidad de Bienestar Universitario, thus complementing the psychotherapeutic process with Computing students. The CRISP-ML(Q) methodology guided the development of the model through six phases: Business and data understanding, data engineering, machine learning model engineering, quality assurance for machine learning applications, model deployment, and monitoring and maintenance. The model achieved 97.28% accuracy in classifying yoga postures in training and validation, and 97% accuracy when it was evaluated with real data. It was integrated into a prototype designed to evaluate the model and conduct the guided yoga routines. The results showed an average reduction of 1.56 points in somatic stress, according to the Hamilton Anxiety Scale, demonstrating a reduction after a single session of use. The model benefited in 66% of the 45 students in Social Work, Clinical Psychology, and Computing subjects who participated in the evaluation phase. The high accuracy of the model and its integration into a graphical interface that provides percentage feedback on performed postures to integrate them with breathing techniques achieve the implementation of therapeutic yoga exercises in Unidad de Bienestar Universitario, benefiting students in reducing their stress levels. Keywords: Yoga, Academic Stress, DenseNet121, Computer Vision |
|---|