Multi-Source Algae Image Detection With Faster RCNN Transfer Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing algae image classification methods are limited by incompatible data formats, require large labeled datasets, struggle with distinguishing difficult species, and lack generalization across different regions and sampling devices, leading to poor classification performance.
Innovation Solution
A method involving automated algae crawling, YOLO v3 target detection, and Faster RCNN transfer learning to create a multi-source algae image target detection model, using a Faster RCNN with MK-MMD for domain adaptation, enabling high-precision detection across varied image sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning-based algae image classification is used, then classification accuracy can be improved, but the requirement for large labeled datasets increases
Solution Approach 1:
The patent applies transfer learning by pre-training a deep neural network on a source domain with abundant labeled data, then adapting it to the target domain with limited labeled data. This preliminary training action on the source domain enables the model to achieve high classification accuracy in the target domain without requiring large amounts of target domain labeled data.
2Device complexity
If algae image classification models are trained on single data source, then model training can be simplified, but data format compatibility decreases
Solution Approach 1:
The patent develops a universal algae image classification model that can process and classify images from multiple data sources including microscope images, IFCB images, and Flowcam images. The model achieves multi-functionality by learning common features across different data formats, enabling it to handle various image types without requiring separate models for each data source.
Solution Approach 2:
The patent uses transfer learning as an intermediary approach to bridge different data sources. By introducing a source domain with abundant data and a target domain with limited data, the transfer learning mechanism enables knowledge transfer between domains, allowing the model to adapt to different data formats while maintaining training efficiency.
3Ease of operation
If traditional algae classification methods are used, then manual identification can be performed, but processing efficiency decreases
Solution Approach 1:
The patent develops an automated algae image classification system that performs classification tasks autonomously without requiring manual identification by experts. The deep neural network model automatically processes algae images, extracts features, and classifies species, enabling the system to serve itself and eliminate the need for time-consuming manual analysis while maintaining high accuracy.
Data Source
AI summary
The present disclosure relates to a method and system for multi-source algae image target detection, and relates to the field of monitoring of algal bloom events in fresh water. The method includes first crawling images of algae of a selected species by using a built automated algae crawling tool, where the images include all formats; classifying and labeling algae in the algae images, and forming a source domain dataset by using all the classified and labeled algae images; performing transfer learning by using a faster recurrent revolutional neural network (Faster RCNN) with reference to a target domain dataset, to obtain a multi-source algae image target detection model; and finally performing identification and classification by using the multi-source algae image target detection model.