Insect Monitoring and Early Detection System for Rice Storage, 2019


Zhongli Pan, adjunct professor, Dept. of Biological and Agricultural engineering, UC Davis

The goal of 2019 research on this project was to further improve a real-time insect monitoring and early detection system for insect activity in rice storage facilities. Engineers previously created and tested a real-time monitoring system consisting of insect traps, USB cameras, LEDs, a tiny computer called a raspberry Pi, a server, and a user interface. Research last year focused on improving handling, convenience, effectiveness, and accuracy, as well as reducing energy use of the imaging system.

The raspberry Pi was replaced with an electronic circuit called a micro-breadboard that was installed in the cap of the insect trap. Wireless Wi-Fi and sensors were added to each trap to measure temperature and relative humidity. AA batteries were also added to provide an independent power source. These changes significantly improved handling, convenience, and energy efficiency of the new system. Also, modifications to the insect-counting algorithm were made to improve accuracy.

The upgraded imaging system consists of the traps, a server, and user interface. A user sends a command to the server, which signals the trap to take a photographic image in the collecting chamber. The user can control how many photos should be taken. Images are then sent back to the server, where they are saved, cropped and processed with the accounting algorithm that counts the number of insects captured per trap. Data related to the number of insects, temperature and relative humidity are then sent back to the user interface.

Setup for laboratory test: (1) clear cylinder, (2) Sample infesting, (3) rice mixing and trap assembling, (4) final setup

The system was evaluated in laboratory experiments and at a commercial rice storage facility. Results were consistent and confirmed high effectiveness and accuracy for monitoring and early detection of insect activity in stored rice. Insect emergence in the laboratory was detected in less than 20 minutes under various infestation levels. After 24 hours, recovery rates ranged between 73% and 83%. Accuracy of image counting in the new system was just over 94%. At the commercial storage facility, it took only 12 minutes to detect insect activity with an accounting accuracy of just over 91%.

This new system could be used for early detection of insect activity in stored rice with high accuracy, reliability, and low costs and labor. However, further research is needed to demonstrate the system for commercial use. The server needs some modification so it can communicate with multiple traps. Apps also need to be developed so users can monitor the system with their smartphones. For scaling up to commercial application, a set of 15 traps need to be tested for insect activity and critical temperature and relative humidity related to insect emergence.