An automatic system for defect detection in plastic crates for glass bottles.
The project describes the design and implementation of an automatic system for detecting defects in plastic crates for glass bottles. In all companies there is damage and defects in their cases, crates, or containers due to constant use, as they are reusable, and therefore this problem causes variou...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2023 |
| País: | Perú |
| Institución: | Universidad Peruana de Ciencias Aplicadas |
| Repositorio: | UPC-Institucional |
| Idioma: | inglés |
| OAI Identifier: | oai:repositorioacademico.upc.edu.pe:10757/673077 |
| Acceso en línea: | https://doi.org/10.1109/CONIITI61170.2023.10324142 http://hdl.handle.net/10757/673077 |
| Access Level: | acceso embargado |
| Palabra clave: | Automated system Image processing Inspection OpenCV Python Raspberry pi TensorFlow https://purl.org/pe-repo/ocde/ford#2.00.00 |
| Sumario: | The project describes the design and implementation of an automatic system for detecting defects in plastic crates for glass bottles. In all companies there is damage and defects in their cases, crates, or containers due to constant use, as they are reusable, and therefore this problem causes various economic losses and a decrease in production, especially in beverage companies. This system was designed to solve and prevent the crates from having defects in their base and containing waste inside, to obtain less product losses in the bottle packaging area. In this research, it is proposed to design the automatic system, which consists of training a convolutional neural network with a database of 136 photographs of waste and defects in the boxes that will be taken by the HQ Raspberry Camera; then programmed into the Raspberry the process of activating the engine so that the box is moved to the point where it will be detected by the photoelectric sensor and the inspection is performed; and finally it is classified indicating whether or not it is in optimal conditions. This is developed in Python using different libraries such as OpenCV, TensorFlow, Tkinter among others. Our results show that the classification and object detection accuracy reached 91.84% out of a bank of 264 tests performed. |
|---|