Related Experiment Video
Updated: May 1, 2026
07:26
In Situ Microscopy for Real-time Determination of Single-cell Morphology in Bioprocesses
Published on: December 5, 2019
7.9K
Investigating quantitative approach for microalgal biomass using deep convolutional neural networks and image
Yang Peng1, Shen Yao1, Aoqiang Li1
1School of Low-Carbon Energy and Power Engineering, China University of Mining and Technology, No 1, Daxue Road, Xuzhou, Jiangsu 221116, China.
Bioresource Technology
|May 26, 2024
Summary
This study introduces an image-based method using convolutional neural networks for monitoring microalgae cultivation. The residual network (RES) model demonstrated superior accuracy and robustness for predicting algae concentration, optimizing energy efficiency.
Area of Science:
- Biotechnology
- Machine Learning
- Aquaculture
Background:
- Effective monitoring of microalgae cultivation is essential for optimizing energy utilization efficiency.
- Image-based analysis offers a promising avenue for non-invasive and quantitative assessment of microalgae biomass.
Purpose of the Study:
- To propose and evaluate a quantitative analysis method for microalgae cultivation monitoring using convolutional neural networks (CNNs).
- To compare the performance of EfficientNet (EFF) and residual network (RES) algorithms for predicting microalgae concentration from images.
Main Methods:
- Developed a quantitative analysis method utilizing microalgae images processed by two CNNs: EfficientNet (EFF) and residual network (RES).
- Tested the method using suspension samples of Rhodophyta (RH) and Spirulina (SP) to simulate real cultivation conditions.
- Assessed prediction accuracy, generalization ability, and robustness of the algorithms.
Main Results:
- The proposed image-based method achieved high prediction accuracy for algae concentration, ranging from 0.94 to 0.99.
- Rhodophyta (RH) showed higher prediction accuracy than Spirulina (SP) due to distinct color shifts.
- The residual network (RES) demonstrated superior generalization and robustness compared to EfficientNet (EFF) and linear regression.
Conclusions:
- The RES algorithm provides a viable and robust approach for image-based quantitative analysis in microalgae cultivation.
- Image analysis using CNNs can effectively monitor microalgae concentration, aiding in process optimization and energy efficiency.
