SRFlow

This implements training of SRFlow on the DIV2K dataset, mainly modified from sanghyun-son/EDSR-PyTorch.

SRFlow Detail

Details, see code/models/SRFlow_model.py

Requirements

• Install PyTorch (pytorch.org)
• pip install -r requirements.txt Note: pillow recommends installing a newer version. If the corresponding torchvision version cannot be installed directly, you can use the source code to install the corresponding version. The source code reference link: Suggestion the pillow is 9.1.0 and the torchvision is 0.6.0

Training

To train a model, run train.py with the desired model architecture and the path to the DIV2K dataset:

# xxx is the decompressed directory of datasets.zip, such as /home/datasets
# 1p train perf
bash test/train_performance_1p.sh --data_path xxx

# 8p train perf
bash test/train_performance_8p.sh --data_path xxx

# 8p train full
# Remarks: Target accuracy 23.05; test accuracy 23.98
# Save the model to experiments/train/models/latest_G.pth
bash test/train_full_8p.sh --data_path xxx 

# 1p eval
# The log file of performance and accuracy is output correctly
bash test/train_eval_1p.sh --data_path xxx 

# finetuning
bash test/train_finetune_1p.sh --data_path xxx 

# online inference demo 
# The prediction results are output correctly, and the output results of multiple runs of the fixed tensor are consistent
python3 demo.py

SRFlow training result

名称	精度	性能	AMP_Type
NPU-1p	-	10.3	O1
NPU-8p	23.98	56.2	O1

Statement

For details about the public address of the code in this repository, you can get from the file public_address_statement.md