Aprendizaje profundo en dispositivo portable para el reconocimiento de frutas y verduras

This project aims to train deep neural networks to recognize around 20 fruits and vegetables using a camera attached to portable devices (Smartphones and embedded systems). We built an acquisition system to gather pictures of different kinds of fruits and vegetables and collected further images from the Internet to train a Convolutional neural network. Instead of defining a new topology and training it from scratch, we took advantage of transfer learning and fine-tuned MobileNet models to classify our images in their corresponding classes. We also trained a lighter model and embedded it both, on a smartphone and on embedded systems (Raspberry pi 3+ and Jetson TX2 development kit). Last but not least, we developed a smartphone application to provide valuable information regarding the fruits and vegetables.

• Autores: Muñoz Bocanegra, Ricardo
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2019
• Institución: Universidad Autónoma de Occidente
• Repositorio: RED: Repositorio Educativo Digital UAO
• Idioma: spa
• OAI Identifier: oai:red.uao.edu.co:10614/11083
• Acceso en línea: http://hdl.handle.net/10614/11083
• Palabra clave: Ingeniería Mecatrónica; Redes neurales (Computadores); Sistemas de computador embebidos; Aplicaciones móviles; Procesamiento de imágenes; Neural networks (Computer science); Embedded computer systems; Mobile apps
• Rights: openAccess
• License: Derechos Reservados - Universidad Autónoma de Occidente