Train the forest using individual samples for each tree

obj.train(imagesFrom, imagesTo, groundTruths)
obj.train(..., 'OptionName',optionValue, ...)

Input

• imagesFrom First sequence of images, a cell array of either filenames or 3-channel color images.
• imagesTo Second sequence of images, same size and format as imagesFrom.
• groundTruth Ground thruth flows, either flow fields or filenames (see cv.readOpticalFlow).

Options

• MaxTreeDepth Maximum tree depth to stop partitioning. default 20
• MinNumberOfSamples Minimum number of samples in the node to stop partitioning. default 3
• DescriptorType Type of descriptors to use. One of:
  • DCT (default) Better quality but slow.
  • WHT Worse quality but much faster.
• PrintProgress Print progress to stdout. default false

Inputs form the training samples (pairs of images and ground truth flows). Sizes of all the provided vectors must be equal. Options are the training parameters.