Multi-Mode Multi-Feature Joint Intelligent Identification Methods for Nematodes

The identification of plant nematodes is crucial in the fields pest control, soil ecology, and biogeography. automated recognition based on deep-learning technology can significantly improve accuracy efficiency their recognition. In this study, we devised a method for multi-mode, multi-feature using techniques which emulated logic domain experts. Beginning with multi-featured nematode dataset, not only designed key feature extraction strategies to address problem weak points small inter-specific differences but also proposed joint training scheme constructed neural network structure interpretability. Finally, an intelligent decision-making expert system was implemented, its performance tested dataset. results indicate that our model achieves up 96.74% identifying 23 species across two-body parts, 17.5% higher than single-part identification. 11 three-body parts reached 98.46%, improvement 1.24% over two-part Our novel demonstrates be increased by incorporating more parts.