| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 7.87 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Esta tese tem como objetivo explorar a aplicação de técnicas de visão 3D na automação de processos logísticos utilizando deep learning para a identificação e manipulação de objetos em ambientes industriais. O trabalho desenvolve um sistema de visão 3D que emprega modelos de deteção de objetos e keypoints treinados com ferramentas como Roboflow e YOLOv8. A metodologia inclui a recolha e anotação de dados, desenvolvimento de modelos de deep learning e análise dos resultados obtidos. Os modelos demonstraram uma precisão média (mAP) de 99.0%, uma precisão de 97.9% e um recall de 96.4% na identificação de blocos. Para a deteção de keypoints, o modelo alcançou um mAP de 98.3%, uma precisão de 96.4% e um
recall de 95.6%. A integração dos modelos apresentou desafios computacionais, mas a abordagem combinada mostrou-se eficaz na identificação precisa de objetos. As limitações incluem a necessidade de otimização de recursos e aprimoramento dos processos de anotação. Futuras pesquisas devem focar-se na deteção da orientação dos blocos para aplicações em braços robóticos.
This thesis aims to explore the application of 3D vision techniques in the automation of logistical processes using deep learning for object identification and manipulation in industrial environments. The work develops a 3D vision system that employs object and keypoint detection models trained with tools such as Roboflow and YOLOv8. The methodology includes data collection and annotation, development of deep learning models, and analysis of the obtained results. The models demonstrated a mean Average Precision (mAP) of 99.0%, a precision of 97.9%, and a recall of 96.4% in block identification. For keypoint detection, the model achieved a mAP of 98.3%, a precision of 96.4%, and a recall of 95.6%. Integrating the models presented computational challenges, but the combined approach proved effective in precise detection. Limitations include the need for resource optimization and improvement in annotation processes. Future research should focus on detecting the orientation of blocks for applications in robotic arms.
This thesis aims to explore the application of 3D vision techniques in the automation of logistical processes using deep learning for object identification and manipulation in industrial environments. The work develops a 3D vision system that employs object and keypoint detection models trained with tools such as Roboflow and YOLOv8. The methodology includes data collection and annotation, development of deep learning models, and analysis of the obtained results. The models demonstrated a mean Average Precision (mAP) of 99.0%, a precision of 97.9%, and a recall of 96.4% in block identification. For keypoint detection, the model achieved a mAP of 98.3%, a precision of 96.4%, and a recall of 95.6%. Integrating the models presented computational challenges, but the combined approach proved effective in precise detection. Limitations include the need for resource optimization and improvement in annotation processes. Future research should focus on detecting the orientation of blocks for applications in robotic arms.
Description
Keywords
3D Vision Logistics Automation Roboflow Object Detection Keypoints YOLOv8 Visão 3D Automação logística Deteção de objetos