A Deep Learning-Based Automatic Mosquito Sensing and Control System for Urban Mosquito Habitats

Area of Science:

• Computer Science
• Entomology
• Public Health

Background:

• Mosquitoes are significant vectors of infectious diseases, necessitating effective control strategies.
• Traditional mosquito control methods can be inefficient and may harm non-target species.

Purpose of the Study:

• To develop and evaluate an automated system for mosquito detection and control using artificial intelligence.
• To assess the efficiency and accuracy of deep learning models in identifying mosquitoes.
• To implement a targeted larvicide delivery system for mosquito population management.

Main Methods:

• Utilized multiple deep learning networks, including Fully Convolutional Network (FCN) and neural network-based regression, for image processing-based mosquito detection.
• Compared the performance of a multi-network system against a single image classifier.
• Implemented an automated larvicide injection system using *Bacillus thuringiensis* israelensis (Bti) in static water bodies.

Main Results:

• The automated system achieved an accuracy of 84% in detecting mosquitoes.
• The single image classifier showed a lower accuracy of 52%.
• Processing time was significantly reduced from 4.64 s to 2.47 s compared to conventional methods.

Conclusions:

• The developed automated system demonstrates high efficiency and accuracy in mosquito detection.
• The integrated approach of AI-driven detection and targeted larvicide application is effective in controlling mosquito proliferation.
• This method offers a more efficient and environmentally conscious alternative to hunting adult mosquitoes, minimizing harm to beneficial insects.