This document provides a brief intro of the usage of builtin command-line tools in derain-toolbox.
For a tutorial that involves actual coding with the API, see our Colab Notebook which covers how to run inference with an existing model, and how to train a builtin model on a custom dataset.
Two scripts in "train.py" and "dist_train.sh" are made to train all the configs provided in this repo. You may want to use it as a reference to write your own training script.
To train a model, first setup the corresponding datasets following dataset preparation, then run:
# single-gpu training
python train.py ${CONFIG_FILE} [optional arguments]
# multi-gpu training
./dist_train.sh ${CONFIG_FILE} ${GPU_NUM} [optional arguments]
Optional arguments are:
--work-dir ${WORK_DIR}
: Override the working directory specified in the config file.--resume-from ${CHECKPOINT_FILE}
: Resume from a previous checkpoint file.--no-validate
: By default, the codebase will perform evaluation every k iterations during the training. To disable this behavior, use--no-validate
--cfg-options
: If specified, the key-value pair optional cfg will be merged into config file.
The evaluate a model's performance, use:
# single-gpu testing
python test.py ${CONFIG_FILE} ${CHECKPOINT_FILE} [optional arguments]
# multi-gpu testing
./dist_test.sh ${CONFIG_FILE} ${CHECKPOINT_FILE} ${GPU_NUM} [optional arguments]
Optional arguments are:
--out
: Specify the filename of the output results in pickle format. If not given, the results will not be saved to a file.--save-path
: Specify the path to store edited images. If not given, the images will not be saved.--seed
: Random seed during testing. This argument is used for fixed results in some tasks such as inpainting.--deterministic
: Related to--seed
, this argument decides whether to set deterministic options for CUDNN backend. If specified, it will settorch.backends.cudnn.deterministic
to True andtorch.backends.cudnn.benchmark
to False.--cfg-options
: If specified, the key-value pair optional cfg will be merged into config file.