options/swinir/train_swinir_sr_classical.json需要修改的内容:
在命令行上运行代码:
python -m torch.distributed.launch --nproc_per_node=1 --master_port=1234 main_train_psnr.py --opt options/swinir/train_swinir_sr_classical.json --dist True
模型保存位置:
下次运行自动从原来的模型基础上进行训练。
使用pycharm命令:
python main_train_psnr.py --opt options/swinir/train_swinir_sr_classical.json
测试性能:
python main_test_swinir.py --task classical_sr --scale 2 --training_patch_size 48 --model_path superresolution/swinir_sr_classical_patch48_x2/models/35000_G.pth --folder_lq testsets/Set5/LR_bicubic/X2 --folder_gt testsets/Set5/HR
--task classical_sr --scale 2 --training_patch_size 48 --model_path superresolution/swinir_sr_classical_patch48_x2/luo/277600_G.pth --folder_lq testsets/Set14/Set14/LRbicx2 --folder_gt testsets/Set14/Set14/GTmod12
由于命名的问题,运行上面命令需要修改:
main_test_swinir.py
img_gt = cv2.imread(path, cv2.IMREAD_COLOR).astype(np.float32) / 255. # img_lq = cv2.imread(f'{args.folder_lq}/{imgname}x{args.scale}{imgext}', cv2.IMREAD_COLOR).astype( # np.float32) / 255. #hujianhua img_lq = cv2.imread(f'{args.folder_lq}/{imgname}{imgext}', cv2.IMREAD_COLOR).astype( np.float32) / 255.