System to measure the position and speed of fish in enclosed environments using image analysis
This study presents the development of a robust system for real-time tracking of guppy fish in closed environments using advanced image segmentation techniques. The primary objectives were to implement precise image segmentation, employ continuous and accurate tracking algorithms, and design an inte...
- Autores:
-
Hernández Vanegas, Rodrigo
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2024
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/74787
- Acceso en línea:
- https://hdl.handle.net/1992/74787
- Palabra clave:
- Convolutional neural network
Image segmentation
Poecilia reticulata
Ingeniería
- Rights
- openAccess
- License
- Attribution 4.0 International
| Summary: | This study presents the development of a robust system for real-time tracking of guppy fish in closed environments using advanced image segmentation techniques. The primary objectives were to implement precise image segmentation, employ continuous and accurate tracking algorithms, and design an interactive user interface for data visualization. Fish, particularly small species like guppy (Poecilia reticulata), serve as excellent models for neurobiological research due to their complex nervous systems and diverse behavioral responses to stimuli. Initial methods, such as optical flow and background subtraction, faced significant challenges due to environmental variations and fish movement. To address these issues, the YOLOv8 (You Only Look Once) convolutional neural network was utilized for its superior accuracy and robustness. The system achieved real-time tracking capabilities with inference speeds around 15 milliseconds per frame on a GPU. The user interface, developed using Flask, HTML, CSS, and JavaScript, effectively visualized the fish’s position and velocity data, allowing for comprehensive behavioral analysis. This system not only enhances tracking accuracy but also provides a reliable tool for neurobiological research, facilitating deeper insights into fish behavior and their responses to stimuli. Future work will focus on optimizing CPU performance and expanding the training dataset to improve the model’s accuracy and generalizability. |
|---|