Multi-dimensional Accuracy Improvement Program
Hardware layer optimization
- Device options: iPhone 12 and above with LiDAR sensors that generate 3D depth maps to aid in identification (especially for thick-bodied insects such as beetles)
- Accessory suggestion: $20 cell phone macro lens clearly captures the setal structure of insects under 0.5cm
Photography Tips
- Best light source: shooting in natural light before 10:00 a.m. to avoid reflective errors caused by midday glare
- Background processing: place solid-color A4 paper as the background to reduce the probability of AI interference (measured accuracy increased by 27%)
Software Functionality Match
- opensMulti-Angle VerificationMode: Take consecutive photos of the dorsal, lateral and ventral surfaces of the insect, and the system will cross-reference the databases
- Use the advanced subscription'sDetail labeling tool: manually circling key features such as tarsus or antennal segment number to aid AI judgment
Note: For Lepidoptera insects, the recognition accuracy is higher in the wing spread state than in the wing retracted state42%
This answer comes from the articleBugPic: a tool for quickly identifying insects by taking photosThe
May not be reproduced without permission:AI productivity tools " How to improve the technical realization of insect recognition accuracy?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil