lumen-ai/com.subhashh.lumen.ai/SmartSearch

SmartSearch

Searching by meaning: the query goes through the model's text half (tens of ms) and is compared with every indexed photo and video frame (a flat scan; a few ms per thousand). A video scores as its best-matching frame. The vectors stay in memory between searches and are reloaded only when the index has changed.

Functions

prepare

suspend fun prepare(context: Context)

Loads the text model and the vectors ahead of time (Search opening), so the first query doesn't wait for them. Safe to call often.

rankedAll

suspend fun rankedAll(context: Context, query: String): List<SmartHit>?

Every item ranked, no cutoff (Search Lab's score curves).

release

suspend fun release()

Frees the models and vectors (Smart Search removed, or memory is tight).

suspend fun search(context: Context, query: String): SmartResult?

Best matches first; empty when nothing clears the bar; null when Smart Search isn't set up. Faces sharpen it (withFaces) when they're downloaded.