Add object detection to the ML.NET CLI

[luisquintanilla]

Add object detection (local) scenario to ML.NET CLI.

[LittleLittleCloud]

@luisquintanilla Here's my proposed CLI command, please review

subcommand

object-detection

available options

Dataset option

• dataset // path to dataset label file, can be coco/vott or tsv(if supported)
• test-dataset // path to test-dataset

Train option

• device // cuda or cpu .default is cpu,
• width // resize width, default is 800
• height // resize height, default is 600
• batch // batch size, default is 20
• epoch // epoch size, default is 10
• seed // default is null
• score-threshold // default is 0.95
• iou-threshold // default is ??

Output option

• name // output project name
• output // output folder

Other option

• verbose
• log-file-path

[LittleLittleCloud]

released in 16.17.17