A utility to load list of paths to training images and annotation files

[imagesPaths, annotationsPaths, success] = cv.Facemark.loadDatasetList(imagesList, annotationsList)

Input

• imagesList The specified file contains paths to the training images.
• annotationsList The specified file contains paths to the training annotations.

Output

• imagesPaths The loaded paths of training images.
• annotationsPaths The loaded paths of annotation files.
• success success flag.

This format is utilized in most facial point annotation datasets (IBUG, HELEN, LPWF, etc.). There are two kinds of files provided in the dataset, images and their corresponding facial point data. The user provides the list of image files and annotations in two separate files. These files can be generated easily using dir or ls commands from the terminal.

The contents of the list files follow a standard format with one path per line.

Example contents of images list file:

/path/to/image1.jpg
/path/to/image2.jpg
...

and contents of corresponding landmarks list file:

/path/to/image1.pts
/path/to/image2.pts
...

where the format of |.pts| files is described in cv.Facemark.loadFacePoints function.