VisionExtractor.hpp File Reference

GPU getter for what VisionAnalyzer found. More...

#include "MayaFlux/Kinesis/Vision/Features.hpp"
#include "MayaFlux/Kinesis/Vision/VisionInsight.hpp"
#include "MayaFlux/Yantra/Executors/TextureExecutionContext.hpp"
#include "MayaFlux/Yantra/Executors/VisionGpuDispatch.hpp"
#include "MayaFlux/Yantra/FunctionalOperation.hpp"

Include dependency graph for VisionExtractor.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	MayaFlux::Yantra::ElementFocus
	How the extractor chooses among several found elements: rank by criterion, highest first when largest, lowest first otherwise. More...
class	MayaFlux::Yantra::VisionExtractor
	Extraction from each VisionAnalysis field. More...

Namespaces
namespace	MayaFlux
	Main namespace for the Maya Flux audio engine.
namespace	MayaFlux::Kakshya
namespace	MayaFlux::Yantra

Enumerations
enum class	MayaFlux::Yantra::ElementCriterion : uint8_t {   MayaFlux::Yantra::Area , MayaFlux::Yantra::Compactness , MayaFlux::Yantra::AspectRatio , MayaFlux::Yantra::Isolation ,   MayaFlux::Yantra::Proximity }
	Per-element measure select_element() ranks find_elements by, taken from what the analyzer already computed: contour area, shape compactness, aspect ratio, nearest-neighbour distance (isolation), or distance from a point. More...
enum class	MayaFlux::Yantra::VisionExtractMode : uint8_t {   MayaFlux::Yantra::Crop , MayaFlux::Yantra::Sample , MayaFlux::Yantra::Patches , MayaFlux::Yantra::Crops ,   MayaFlux::Yantra::SampleRegions , MayaFlux::Yantra::SamplePoints , MayaFlux::Yantra::Silhouettes , MayaFlux::Yantra::Equalize ,   MayaFlux::Yantra::Expose , MayaFlux::Yantra::Edges , MayaFlux::Yantra::MotionMask , MayaFlux::Yantra::Annotate ,   MayaFlux::Yantra::SelectElement }
	Which method run_operation dispatches to. More...

Detailed Description

GPU getter for what VisionAnalyzer found.

Every method takes a Kinesis::Vision::VisionAnalysis and derives its geometry from whichever field is populated; an analysis with nothing in it extracts nothing. All pixel work runs on the GPU, the host only packs the analysis into buffers and dispatches.

FunctionalOperation<shared_ptr<SignalSourceContainer>, vector<DataVariant>>, like VisionAnalyzer. run_operation reads the analysis from Datum metadata key "vision_analysis" (the key VisionAnalyzer::run_operation writes) and runs the method selected by VisionExtractMode, so a VisionAnalyzer -> VisionExtractor chain needs no adapter. mask() alone stays outside that path: it needs the VisionGpuExecutor that owns the label buffer.

Definition in file VisionExtractor.hpp.