The three recognition modes
KORENANI has verified recognition handling for general, insect, and plant modes. The API accepts those annotation values, and the iOS implementation also maps those modes for recognition flows. Fact refs: docs/seo/02-facts.md:18-30.
General
General recognition is the broad object-recognition path. The backend general image-processing flow calls the image-analysis port. Fact refs: docs/seo/02-facts.md:18, docs/seo/02-facts.md:24, docs/seo/02-facts.md:40.
Insect
Insect recognition is handled as a distinct recognition path. The backend insect image-processing flow calls the insect annotation port. Fact refs: docs/seo/02-facts.md:19, docs/seo/02-facts.md:25, docs/seo/02-facts.md:42.
Plant
Plant recognition is also handled as a distinct recognition path. The backend plant image-processing flow calls the plant annotation port. Fact refs: docs/seo/02-facts.md:20, docs/seo/02-facts.md:26, docs/seo/02-facts.md:41.
Manual and keyword paths
The codebase also contains manual generation and keyword-related paths. For public feature copy, this site focuses on the current general, insect, and plant explanation and avoids presenting deprecated subscription labels as current user-facing modes. Fact refs: docs/seo/02-facts.md:37-39, docs/seo/02-facts.md:245.
Claims to avoid
This page does not say that recognition is processed only on the device. Non-manual recognition sends image data to the API. For more detail, read How Photo Data Flows During Image Recognition.