Loading...
Please wait, while we are loading the content...
Similar Documents
Aplicación móvil para el prototipado de interfaz gráfica de usuario en la plataforma android mediante el reconocimiento de bocetos utilizando visión artificial.
| Content Provider | Semantic Scholar |
|---|---|
| Author | Espinoza, Aguilar Luis, José |
| Copyright Year | 2019 |
| Abstract | Amateur or beginner developers of mobile applications face the creation of prototypes of graphical user interface in their projects. In order to run this process, they are forced to search and get training about the tools that are aimed to Graphic Design professionals; distracting the time allocated for the analysis and development of Software. The objective of this study is to determine whether artificial vision facilitates the creation of prototypes of graphical user interface, through an application, which is composed of a deep learning server (Machine Learning - Engine), a database server and functions on the cloud (Firebase) and a mobile app for smartphones with Android operating system. Such application allows to capture a photo of a sketch drawn by the user; then, it is sent to the deep learning server to detect the possible components that have been drawn. Depending on the result, it is generated the graphical interface prototype at the location and type detected by the artificial vision; then, it will be managed with the available options within the mobile app. For the artificial vision it was used a pre-trained model of convolutional neural network Faster R-CNN, trained and evaluated with TensorFlow library and deployed on IBM Power AI Vision and Machine Learning - Engine servers. For the training, a set of hand strokes collected from 19 students was used, which were treated using distortion techniques using OpenCV, in order to increase the data to a total of 862 images, and thus improve the detection quality of the model. Finally, a test scenario was developed with 19 students from the Universidad Nacional de Loja of the System Engineering Career, where 95% positively affirmed the operation and purpose of the application, validating the correct running of the system and of this degree work; however, the application can be improved by increasing the quantity and quality of images in the data set, and adding new graphic components to the Design Material guide for Android; this would improve the quality of detection and diversity of components by the artificial vision model. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://dspace.unl.edu.ec/jspui/bitstream/123456789/21989/1/Aguilar%20Espinoza,%20Jos%C3%A9%20Luis.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
