Back to Adaptive Vision Library website

You are here: Start » Function Reference » Deep Learning » DeepLearning_DetectFeatures


Performs feature classification using trained deep-learning model.



void avl::DeepLearning_DetectFeatures
	DeepLearningConnectionState& ioState,
	const avl::Image& inImage,
	const atl::Optional<avl::Region>& inRoi,
	const avl::DeepModel_FeatureDetection& inDeepModel,
	const bool& inOverlap,
	atl::Array<avl::Image>& outHeatmaps,
	atl::Optional<avl::Image&> outFeature1 = atl::NIL,
	atl::Optional<avl::Image&> outFeature2 = atl::NIL,
	atl::Optional<avl::Image&> outFeature3 = atl::NIL,
	atl::Optional<avl::Image&> outFeature4 = atl::NIL


Name Type Default Description
ioState DeepLearningConnectionState& Object used to maintain state of the function.
inImage const Image& Input image
inRoi const Optional<Region>& NIL Area of interest
inDeepModel const DeepModel_FeatureDetection& Trained model
inOverlap const bool& True Add tiles overlapping to improve results quality
outHeatmaps Array<Image>& Returns heatmaps for all classes
outFeature1 Optional<Image&> NIL Returns heatmap for first feature class
outFeature2 Optional<Image&> NIL Returns heatmap for second feature class or empty image if class is not specified
outFeature3 Optional<Image&> NIL Returns heatmap for third feature class or empty image if class is not specified
outFeature4 Optional<Image&> NIL Returns heatmap for fourth feature class or empty image if class is not specified


For input inImage only pixel formats are supported: 1xuint8, 3xuint8.

Read more about pixel formats in Image documentation.


Loading Deep Learning Model may take longer timer. Consider using DeepLearning_LoadModel for pre-loading model before execution starts.


  • Model provided on inDeepModel input will be loaded to service automatically on first usage of Deep Learning filters.
  • Service automatically releases and loads model into its memory. It may result in releasing model previously loaded using DeepLearning_LoadModel filter.
  • These filters only communicates with service and cannot be use for parallel computation.


Error type Description
DomainError Empty ROI region in DeepLearning_DetectFeatures
DomainError Size of ROI differs from size of image in DeepLearning_DetectFeatures
DomainError Invalid DeepModel in DeepLearning_DetectFeatures.
DomainError Empty image in DeepLearning_DetectFeatures.

See Also

  • Models for Deep Learning may be created using Adaptive Vision Deep Learning Editor or using Training Api.