Automated Algae Detection using Machine Learning
Based on drone photos from Skive Fjord, we have build up our first machine learning model which can detect sea lettuce bloom.
The distinctive color of algae has been used to develop computer vision-based algae monitoring systems. However, traditional computer vision pipelines do not have high repeatability because they dependent on the effectiveness of the feature detectors or the segmentation procedure, which can be ineffective by fluctuating illumination, occlusion or the presence of comparable objects in the background. Classification algorithms, reflectance band-ratio algorithms and spectral band difference algorithms take the spectral data as input to detect the presence of algae in bodies of water. Although these algorithms have been successful in monitoring micro algal blooms in the open ocean, they have not been validated with macro algae.
With the use of Machine Learning, an algae monitoring system can be made which is robust to changes in image parameters (e.g., image size, resolution, orientation). This system can be adjusted to the environmental conditions and algae species under consideration on a global scale and is able to work with a wide variety of robot platforms.