CLI Options

DarkMark cli options:

Command	Examples	Description
add=<path>	add=/home/bob/nn/cars	Add the specified directory as a new project, then load that project as if load=... had been specified.
batch_size=<number>	batch_size=64	The batch size to use when generating the Darknet .cfg file.
class_imbalance=<bool>	class_imbalance=false	Toggles the counter_per_class setting when generating the Darknet .cfg file.
cutmix=<bool>	cutmix=false	Determines if image cutmix is enabled.
darknet=run	darknet=run	Auto-run the darknet generation when combined with editor=gen-darknet.
del=<path>	del=/home/bob/nn/cars	Delete the project that matches the specified directory. This does not delete the files, only the project definition.
do_not_resize_images=<bool>	do_not_resize_images=true	Determines if images are left "as-is". See Do Not Resize Images.
editor=<name>	editor=gen-darknet	Action to perform from the main editor window. Only value supported is gen-darknet.
flip=<bool>	flip=false	Enable horizontal image flip.
height=<number>	height=416	Network dimensions to use when generating the Darknet .cfg file.
learning_rate=<number>	learning_rate=0.001	The learning rate to use when generating the Darknet .cfg file.
limit_neg_samples=<bool>	limit_neg_samples=true	Determines if negative samples should be limited.
limit_validation_images=<bool>	limit_validation_images=true	Determines if validation images should be limited.
load=<name>	load=cars load=123456 load=/home/username/nn/cars	Hide the DarkMark Launcher and use the given project instead. Can use the project name, the project key, or the directory. The project must already exist in DarkMark.
max_batches=<number>	max_batches=20000	The number of iterations to use when generating the Darknet .cfg file.
mixup=<bool>	mixup=false	Determines if image mixup is enabled.
mosaic=<bool>	mosaic=false	Determines if image mosaic is enabled.
resize_images=<bool>	resize_images=true	Determines if images are resized to match the network dimensions. See Resize Images.
restart_training=<bool>	restart_training=false	Determines if training should restart with the previous existing weights (when set to true) or start from scratch (when set to false).
subdivisions=<number>	subdivisions=2	The number of subdivisions to use when generating the Darknet .cfg file.
template=<filename>	template=/home/bob/src/darknet/cfg/yolov4-tiny.cfg	Configuration template to use when combined with load=...
tile_images=<bool>	tile_images=true	Determines if image tiling should be enabled. See Tile Images.
width=<number>	width=416	Network dimensions to use when generating the Darknet .cfg file.
yolo_anchors=<bool>	yolo_anchors=true	Determines if the YOLO anchors should be re-calculated.
zoom_images=<bool>	zoom_images=true	Determines if crop-and-zoom is enabled when processing images. See Crop & Zoom Images.

Boolean flags can be set using true, false, yes, no, on, off, 0, or 1.

Examples, assuming there is a project called "animals":

DarkMark load=animals editor=gen-darknet darknet=run

Or:

DarkMark load=animals editor=gen-darknet

Or:

DarkMark load=animals tile_images=on limit_neg_samples=off max_batches=10000 width=416 hight=416 editor=gen-darknet

Or:

DarkMark add=/home/bob/nn/animals editor=gen-darknet width=608 height=608 do_not_resize=true darknet=run

Or:

DarkMark del=/home/bob/nn/animals