Skip to content
This repository has been archived by the owner on May 9, 2024. It is now read-only.

Commit

Permalink
Modelmaker integration v1
Browse files Browse the repository at this point in the history
  • Loading branch information
Badrinath Gonnabattula authored and Manu Mathew committed May 10, 2023
1 parent 185c37b commit f108240
Show file tree
Hide file tree
Showing 5 changed files with 220 additions and 182 deletions.
4 changes: 4 additions & 0 deletions references/edgeailite/engine/train_pixel2pixel.py
Expand Up @@ -91,14 +91,18 @@ def get_config():
args.dataset_name = 'cityscapes_segmentation' # dataset type
args.transforms = None # the transforms itself can be given from outside
args.input_channel_reverse = False # reverse input channels, for example RGB to BGR
args.annotation_prefix = None # Annotations prefix name

args.data_path = './data/cityscapes' # 'path to dataset'
args.save_path = None # checkpoints save path
args.phase = 'training' # training/calibration/validation
args.date = None # date to add to save path. if this is None, current date will be added.
args.output_dir = None

args.logger = None # logger stream to output into
args.show_gpu_usage = False # Shows gpu usage at the begining of each training epoch
args.device = None
args.distributed = None

args.split_file = None # train_val split file
args.split_files = None # split list files. eg: train.txt val.txt
Expand Down

0 comments on commit f108240

Please sign in to comment.