Related Experiment Video
Updated: Nov 12, 2025
08:20
Author Spotlight: AI-Driven Trypanosome Species Detection from Microscopic Images
Published on: October 27, 2023
2.1K
Recognizing and counting Dendrocephalus brasiliensis (Crustacea: Anostraca) cysts using deep learning.
Angelica Christina Melo Nunes Astolfi1, Gilberto Astolfi2,3, Maria Gabriela Alves Ferreira1
1Faculty of Engineering, Architecture and Urbanism, and Geography, Federal University of Mato Grosso do Sul, Campo Grande, MS, Brazil.
Plos One
|March 18, 2021
Summary
Automated detection and counting of Dendrocephalus brasiliensis cysts using YOLOv3 is efficient. This method surpasses manual counting and Faster R-CNN, offering accurate results for conservation and production activities.
Area of Science:
- Aquaculture and Conservation Biology
- Biotechnology and Image Analysis
- Crustacean Biology
Background:
- Dendrocephalus brasiliensis is a South American freshwater crustacean vital for aquaculture and conservation.
- Accurate counting of its resistant cysts is crucial for research and production but challenging manually.
- Manual cyst counting is laborious, time-consuming, and prone to errors.
Purpose of the Study:
- To develop and validate an automated method for detecting and counting Dendrocephalus brasiliensis cysts.
- To compare the performance of YOLOv3 and Faster R-CNN object detection models for this task.
- To establish a new dataset for training and evaluating automated cyst detection algorithms.
Main Methods:
- Creation of the DBrasiliensis dataset, comprising 246 images with 5141 Dendrocephalus brasiliensis cysts.
- Training and evaluation of YOLOv3 (You Only Look Once) and Faster R-CNN (Region-based Convolutional Neural Networks) object detection models.
- Comparative analysis of model accuracy, R2, Root Mean Square Error (RMSE), and Mean Absolute Error (MAE) for cyst detection and counting.
Main Results:
- YOLOv3 demonstrated superior performance over Faster R-CNN in detecting and counting Dendrocephalus brasiliensis cysts.
- YOLOv3 achieved an accuracy rate of 83.74%, R2 of 0.88, RMSE of 3.49, and MAE of 2.24.
- The study confirmed the feasibility of inferring total cyst counts from automated sample analysis.
Conclusions:
- The automated YOLOv3 approach is effective and accurate for detecting and counting Dendrocephalus brasiliensis cysts.
- This method offers a significant improvement over manual counting, enhancing efficiency in research and aquaculture.
- The DBrasiliensis dataset provides a valuable resource for future studies in automated crustacean cyst analysis.
