Progress 09/01/09 to 08/31/12
Outputs OUTPUTS: Activities: Collection and analysis of honey bee recordings including colonies infected with hive beetles, foul brood, Africantized, CCD, and Ceranea. Electronics development in a USB based ultrasonic microphone. Events: Demonstrated the new bee scanner device at the Eastern Apicultural Society meeting in August 2012, and the Montana State Beekeepers meeting in October 2012. Products: Produced a fully operational handheld device for the rapid evaluation of honey bee colony health using sonic fingerprints. A byproduct of this device is a simple USB ultrasonic microphone with potential applications in other fields of study. PARTICIPANTS: Robert Seccomb - Principal Investigator Jerry Bromenshenk - Bee Researcher and Public Outreach Scott Debnam - Senior Bee Researcher Will Leishman - Electronics Technician Joshua Rice - Electronic Technician Lupine Logic, Joe Glassy - Subcontracted computing platform design TARGET AUDIENCES: Honey bee researchers, commercial beekeepers and hobbyist beekeepers. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.
Impacts Initial use of a single Artificial Neural Network to evaluate bee sounds for a number of maladies proved unsuccessful. Use of an ensemble of networks, each trained to a specific condition, improved accuracy of diagnosis from 86% to 92% correct. Advances in electronics provided an inexpensive alternative to existing computing platforms, allowing us to develop a fully operational device in short amount of time.
Publications
- Sonic Analysis for Rapid Detection of Varroa Mites and Other Pathologies without Opening the Beehive, Interim Report to USDA NIFA, March 2011, R Seccomb
- Sonic Analysis for Rapid Detection of Varroa Mites and Other Pathologies without Opening the Beehive, Final Report to USDA NIFA, October 2012, R Seccomb
|
Progress 09/01/09 to 08/31/10
Outputs OUTPUTS: Collected diseased and health bee colony recordings, including Africanized colonies from the SouthWestern US and Mite free colonies from Australia. Processed recordings using existing artificial neural networks system, and alternate analysis methods, including ensemble networks to increase accuracy on multiple bee diseases. Produced preliminary schematics for a dedicated recording and analysis device. PARTICIPANTS: Individuals Robert Seccomb - PI, programmer; Jerry Bromenshenk - Co-PI, bee expert; Scott Debnam - Head bee wrangler; Sarah Red-Laird - bee technician; Robert Etter - electronics design; Josh Rice - electronics technician; Organizations Lupine Logic; Collaborators Bee keepers throughout the US, Canada, and Australia TARGET AUDIENCES: Nothing significant to report during this reporting period. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.
Impacts Use of an ensemble (multiple) artificial neural networks appears the correct approach for analysising a multitude of bee diseases, rather than a single network. Use of a greater audio spectrum is also indicated, and use of ultra-sonics is worth investigating.
Publications
- No publications reported this period
|
|