Algomox-Super_Resolution

Dataset folder contains the dataset used for both training and testing

SRGAN_HPT

Is for Hyper Parameter Tuning

command : nnictl create --config path\to\config.yaml

Only a sample dataset is used for this task

Parameter range and choices can be adjusted in the search space

SRGAN_NAS

Is for Neural Architecture Search

command : nnictl create --config path\to\config.yaml

Only a sample dataset is used for this task

Parameter range and choices can be adjusted in the search space

SRGAN_Training

Contains files for training (both pre-training and finetuning), validation and testing

Training command : !python main.py --LR_path SRGAN_CustomDataset\custom_dataset\train_LR --GT_path SRGAN_CustomDataset\custom_dataset\train_HR

Validating Command : !python main.py main.py --mode valid --LR_path Dataset\valid\LowRes --GT_path Dataset\valid\HighRes --generator_path SRGAN_CustomDataset\model\pre_trained_model_2700.pt

Testing Command !python main.py main.py --mode test --LR_path Dataset\valid\LowRes --GT_path Dataset\valid\HighRes --generator_path SRGAN_CustomDataset\model\pre_trained_model_2700.pt

Check the available command line arguments in main.py

Python Scripts for Data Pre-processing

Augment_RC.py

downsample.py

GrayScaleRM.py