A Brazilian native bee (Tetragonisca angustula) dataset for computer vision

Area of Science:

• Ecology and Environmental Science
• Computer Science and Artificial Intelligence
• Agricultural Science

Background:

• Jataí bees are essential pollinators, making their sustainable management critical for ecosystem services.
• Precision agriculture benefits from effective pollinator management and monitoring.
• Current monitoring methods for bee populations can be labor-intensive and lack scalability.

Purpose of the Study:

• To create a comprehensive dataset of Jataí bee activity for computer vision applications.
• To facilitate the development of intelligent monitoring systems for natural and artificial bee hives.
• To support research in deep learning models for bee behavior analysis and population assessment.

Main Methods:

• Acquired video data from natural and artificial Jataí bee hives in diverse environments.
• Positioned cameras at hive entrances to capture bee traffic.
• Extracted images from videos and created labeled data for training bee detection models.

Main Results:

• Developed a novel dataset comprising videos, images, labels, and metadata of Jataí bee colonies.
• The dataset is suitable for training and evaluating computer vision models for bee tracking and detection.
• Demonstrated the potential for using this dataset in comparative studies of deep learning algorithms.

Conclusions:

• The created dataset is a valuable resource for advancing computer vision in apiology.
• This resource can significantly contribute to the development of intelligent, automated systems for monitoring Jataí bee populations.
• Facilitates research into sustainable beekeeping practices and enhances precision agriculture through improved ecosystem service monitoring.