The application adopts patented compression technology to control the core database volume at 38MB, realizing the local storage of 2000 common insect characteristics. The data architecture adopts a layered design: base layer (morphological feature vectors), extension layer (ecological behavioral coding), and emergency layer (hazardous treatment scheme). This design allows the recognition accuracy to remain 78% when there is no network.
Intelligent synchronization mechanism will automatically update the data of species in hotspot areas when there is network, for example, when users are in the Yunnan rainforest, they will be given priority to download the data of Anopheles family. The data is encrypted with military-grade AES-256 standard, and the collection records can be exported to Darwin Core format, which is commonly used in scientific research.
Tests show that the last 30 recognition records can still be called stably in extreme environments such as Everest Base Camp. The technology solution won the 2023 Mobile Innovation Award and will soon open source the database compression algorithm.
This answer comes from the articleBugPic: a tool for quickly identifying insects by taking photosThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil