Ayuda
Ir al contenido

Dialnet


Métodos objetivos estadísticos de valoración de la magnitud de interés en base a las trayectorias de dirección de visionado

  • Autores: Manuel López Palma
  • Directores de la Tesis: Javier Gago Barrio (dir. tes.), Montserrat Corbalan Fuertes (codir. tes.)
  • Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2020
  • Idioma: español
  • Materias:
  • Texto completo no disponible (Saber más ...)
  • Resumen
    • In this work, a new method is presented to quantify the attention paid by people to the different objects in their environment.

      The new metric models the eye to determine the attention given and quantifies that attention at all points in an area of interest. This metric is compared with the current one based on time and it is justified that ours is more in line with human perception.

      For the calculation of said attention, the concept of oriented trajectory is introduced as the set of positions and orientation angles of the head, of each person of interest and in the time that is of interest. We will justify that only with this data can such care be determined.

      In the presented method, top view cameras are used as a system to have the highest performance with the minimum number of cameras. Likewise, two methods are analyzed: the 3D method that uses depth information, and a less precise method, 2D, that only uses imaging cameras.

      This thesis also presents a way of calculating the time metric, a method that is widely used today to verify how many people and in how long an ad has been seen. The form presented by our method allows reducing the number of cameras required, and therefore it is advantageous in terms of the resources required for its implementation.

      Finally, the results are verified using a camera in the front part of the head simulating the eye and an IMU sensor that measures the angles of the head. In this way, the attention relationship of the objects detected by the camera is determined, and the same attention relationship of the objects obtained by the proposed method.


Fundación Dialnet

Dialnet Plus

  • Más información sobre Dialnet Plus

Opciones de compartir

Opciones de entorno