JeVoisBase component algorithms

JeVois provides the ability to implement individual machine vision algorithms as Component objects, which combine the implementation of the algorithm itself, and of parameters that are exposed to users to tune the algorithm's operation. More...

Collaboration diagram for JeVoisBase component algorithms:

Modules
	Image filters
	Collection of filters using OpenCV filtering functions.
	JeVois profiler for Darknet deep neural networks
	These functions simply establish a C linkage to jevois::Profiler, to be used to profile the forward pass in Darknet networks so that layer sizes and network structure can be optimized.

Classes
class	ArUco
	Simple wrapper class over the opencv_contrib ArUco augmented reality markers. More...
class	EyeTracker
	Eye-tracker class used to detect gaze direction from close-up video of one's eye. More...
class	FaceDetector
	Face detection using OpenCV. More...
class	FilterGPU
	Simple image filtering using OpenGL-ES on the GPU. More...
class	SuperPixel
	Superpixel image segmentation from OpenCV. More...
class	BlobDetector
	Simple color-based object/blob detection/tracking. More...
class	ObjectMatcher
	Object matching using OpenCV keypoint detection and matching. More...
class	ObjectRecognitionBase
	Abstract base class for an object recognition component. More...
class	ObjectRecognition< NetType >
	Wrapper around a neural network implemented by with the tiny-dnn framework by Taiga Nomi. More...
class	ObjectRecognitionCIFAR
	Object recognition CNN for CIFAR-10 dataset. More...
class	ObjectRecognitionILAB
	Object recognition CNN for ILAB-10 dataset. More...
class	ObjectRecognitionMNIST
	Object recognition CNN for MNIST handwritten digits. More...
class	FastOpticalFlow
	Fast optical flow computation using dense inverse search. More...
class	QRcode
	QRcode and Barcode detection using ZBar. More...
class	RoadFinder
	Navigation by finding road. More...
class	Saliency
	Simple wrapper class around Rob Peter's C-optimized, fixed-point-math visual saliency code. More...
class	Surprise
	Compute Itti & Baldi surprise over video frames. More...
class	Kalman1D
	Simple component to track a moving 2D point over time using a Kalman filter. More...
class	Kalman2D
	Simple component to track a moving 2D point over time using a Kalman filter. More...
class	BufferedVideoReader
	Simple class to read video frames from a movie file, decode them, and buffer them for smooth playback. More...

Detailed Description

JeVois provides the ability to implement individual machine vision algorithms as Component objects, which combine the implementation of the algorithm itself, and of parameters that are exposed to users to tune the algorithm's operation.

Component algorithms can be shared and assembled into hierarchies to give rise to different machine vision modules.