NTIRE 2023 视频超分辨率赛道
学习

NTIRE 2023 视频超分辨率赛道

1 Clone our repo git clone <https://github.com/yindaheng98/NTIRE2023_ESR> Our model path is: NTIRE2023_ESR/model_zoo/PRFDN_28.992.pth Our network path

1 Clone our repo

git clone <https://github.com/yindaheng98/NTIRE2023_ESR>

Our model path is: NTIRE2023_ESR/model_zoo/PRFDN_28.992.pth

Our network path is NTIRE2023_ESR/models/rfdn_half

2 Test our code

Execute the following script to set the ROOT variable in the shell to the project path(e.g. ROOT=/home/seu/NTIRE2023_ESR) and create the directory results to save the results.

ROOT=PATH_TO_OUR_PROJECT
rm -rf "$ROOT/results"
mkdir -p "$ROOT/results"

You can select the GPU by setting CUDA_VISIBLE_DEVICES=[GPU_id](e.g CUDA_VISIBLE_DEVICES=0).

Then you can execute our test code as follow:

python test_demo.py \\
  --data_dir PATH_TO_DATASET \\
  --save_dir "$ROOT/results" \\
  --model_id 13

The directory structure of the dataset of PATH_TO_DATASET is the same as that of the official repository. Our model id is 13.

The results will be stored in ROOT/results.json and ROOT/results.txt.

To check the results you can execute the following script:

printf "%20s %12s %17s %14s %5s\\n" model_name valid_memory valid_ave_runtime valid_ave_psnr flops
for line in $(cat results.json | jq -r 'to_entries|.[]|[.key,.value.valid_memory,.value.valid_ave_runtime,.value.valid_ave_psnr,.value.flops|tostring] | join(",")'); do
  printf "%20s %12f %17f %14f %5f\\n" $(echo $line | sed 's/,/ /g')
done

The results then will be displayed.

Comment