Figure 1
Citizen scientist-AI data ecosystem mediated through the GLOBE Observer data collection mobile app. Human touch elements in the data procurement and analysis system are indicated in red. Blue arrows describe data pathways between user inputs in the app, AI routines, and data return. Dashed outlines indicate processes that are planned or in development. (a) Data collection app. (b) manual image review. (c) GLOBE database. (d) citizen scientist entry points to GLOBE science. (e) Geofence capability alerts volunteers for project-specific targeted data collection requests submitted by community users. (f) Citizen scientist tasks contributing to AI development and classification processes. (g) Data analysis outcomes. The AI-mediated data processes include (1) cloud database ingest of submitted data, (2) data extraction from submitted citizen scientist images, and (3) AI-driven data collector feedback and data enrichment through GeoAI integration.
Figure 2
Examples of photos flagged or blurred using AI screening: (a) license plate text; (b) face; (c) feet captured in a downward photo; (d) useful text data lost from AI blurring of features; (e) classification error where the AI system recognized mosquito larvae as text.
Figure 3
Screenshots of (a) the GLOBE Observer mobile app landing screen, (b) the drop-down menu listing selectable breeding habitats, and (c) the identification key built into the app to assist citizen scientists with identification of the medically important mosquito genera.
Figure 4
Localization test result of an Anopheles stephensi specimen: (a) full larva view; (b) boxes identifying segmentation of image for analysis; and (c) labeled data segments.
Figure 5
GLOBE Observer land cover tool data collection protocol. (a) Examples of four directional views at each location. (b) Images obtained from the four cardinal directions for a series of land cover types (from Huang et al. 2023, reproduced with permission of the author).