Bibliographic citations
This is an automatically generated citacion. Modify it if you see fit
Contreras, S., Gallegos, F. (2024). TryOut: Generación de pruebas de ropa a partir de fotos [Tesis, Universidad Peruana de Ciencias Aplicadas (UPC)]. http://hdl.handle.net/10757/676201
Contreras, S., Gallegos, F. TryOut: Generación de pruebas de ropa a partir de fotos [Tesis]. PE: Universidad Peruana de Ciencias Aplicadas (UPC); 2024. http://hdl.handle.net/10757/676201
@misc{renati/892809,
title = "TryOut: Generación de pruebas de ropa a partir de fotos",
author = "Gallegos Quispe, Franco David",
publisher = "Universidad Peruana de Ciencias Aplicadas (UPC)",
year = "2024"
}
Title: TryOut: Generación de pruebas de ropa a partir de fotos
Other Titles: TRYOUT: GENERATING CLOTHING TESTS FROM PHOTOS
Advisor(s): Reyes Silva, Patricia Daniela
Keywords: Visión computacional; Modelos virtual try on; Prueba de ropa virtual; Modelos de super resolución; Experiencia de compra online; Computational Vision; Virtual Try On Models; Virtual clothes fitting; Super resolution models; Online shopping experience
OCDE field: https://purl.org/pe-repo/ocde/ford#2.02.04; https://purl.org/pe-repo/ocde/ford#1.00.00
Issue Date: 20-Mar-2024
Institution: Universidad Peruana de Ciencias Aplicadas (UPC)
Abstract: La venta en línea de prendas experimentó un notorio aumento en su crecimiento a partir del año 2020. No obstante, en este escenario, los usuarios se enfrentan a un desafío al momento de realizar sus compras. Específicamente, los usuarios experimentan una sensación de inseguridad debido a la falta de la opción de probarse las prendas antes de concretar la compra.
El presente proyecto propone una forma de que los usuarios puedan probarse las prendas de manera online. De esta manera, se definió como objetivo desarrollar una aplicación web que permita a los usuarios probarse ropa de manera virtual. Para ello, se realizó una búsqueda de modelos VTO (Virtual Try On) que permiten generar imágenes de prueba de ropa. Se seleccionó el modelo VTO a partir de métricas de interés como el tiempo de inferencia y métricas que cuantifican la calidad de las imágenes generadas. Asimismo, se encontró limitaciones a estos modelos, los cuales la mayoría generaban imágenes de tamaño pequeño, en específico, 192 por 256 pixeles. Por lo que, se propuso a investigar sobre métodos que puedan aumentar el tamaño de la imagen sin perder significativamente la calidad de estas.
A partir de ello, se desarrolló una API del modelo VTO para que pueda ser consumido por la aplicación web desarrollada y realizar una encuesta a un grupo de usuarios para evaluar el desempeño del modelo en un entorno real. Los resultados de la experiencia del modelo son positivos, ya que los usuarios muestran seguridad respecto a las imágenes generadas.
De esta manera, se propone una solución a la necesidad que tiene el usuario de probarse las prendas de manera online antes de realizar la compra de éstas.
The online sale of garments experienced a notorious increase in its growth as of 2020. However, in this scenario, users face a challenge when making purchases. Specifically, users experience a sense of insecurity due to the lack of the option to try on garments before making the purchase. This project proposes a way for users to try on clothes online. Thus, the objective was defined as the development of a web application that allows users to try on clothes virtually. To this end, a search was carried out for VTO (Virtual Try On) models that allow the generation of images for trying on clothes. The VTO model was selected based on metrics of interest such as inference time and metrics that quantify the quality of the generated images. Also, limitations were found to these models, which most of them generated images of small size, specifically, 192 by 256 pixels. Therefore, it was proposed to investigate methods that could increase the size of the image without significantly losing the quality of these. From this, an API of the VTO model was developed so that it can be consumed by the developed web application and a survey was conducted to a group of users to evaluate the performance of the model in a real environment. The results of the model experience are positive, as users show confidence regarding the generated images. In this way, a solution is proposed to the user's need to try on garments online before purchasing them.
The online sale of garments experienced a notorious increase in its growth as of 2020. However, in this scenario, users face a challenge when making purchases. Specifically, users experience a sense of insecurity due to the lack of the option to try on garments before making the purchase. This project proposes a way for users to try on clothes online. Thus, the objective was defined as the development of a web application that allows users to try on clothes virtually. To this end, a search was carried out for VTO (Virtual Try On) models that allow the generation of images for trying on clothes. The VTO model was selected based on metrics of interest such as inference time and metrics that quantify the quality of the generated images. Also, limitations were found to these models, which most of them generated images of small size, specifically, 192 by 256 pixels. Therefore, it was proposed to investigate methods that could increase the size of the image without significantly losing the quality of these. From this, an API of the VTO model was developed so that it can be consumed by the developed web application and a survey was conducted to a group of users to evaluate the performance of the model in a real environment. The results of the model experience are positive, as users show confidence regarding the generated images. In this way, a solution is proposed to the user's need to try on garments online before purchasing them.
Link to repository: http://hdl.handle.net/10757/676201
Discipline: Ciencias de la Computación
Grade or title grantor: Universidad Peruana de Ciencias Aplicadas (UPC). Facultad de Ingeniería
Grade or title: Licenciado en Ciencias de la Computación
Juror: Diaz Suarez, Jorge Eduardo; Rosales Huamanchumo, Javier; Ubaldo Gamarra, Victoria
Register date: 22-Oct-2024
This item is licensed under a Creative Commons License