GramSR/inference.py at main · aimagelab/GramSR · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
import os
import argparse
import torch
from torchvision import transforms
from tqdm import tqdm

from gramsr import GramSR
from src.my_utils.wavelet_color_fix import adain_color_fix, wavelet_color_fix
from src.datasets.dataset import TestDatasetv2


def main(args):
    pred_dir = os.path.join(args.output_dir, f"pred")
    gt_dir = os.path.join(args.output_dir, f"GT")

    os.makedirs(pred_dir, exist_ok=True)
    os.makedirs(gt_dir, exist_ok=True)

    print(f"[INFO] Pred dir: {pred_dir}")
    print(f"[INFO] GT dir:   {gt_dir}")

    model = GramSR(args)

    model.add_dino()
    model.load_dino(args.lora_dir, 53001)
    model.unet.set_adapter([
        'default_encoder_pix', 'default_decoder_pix', 'default_others_pix',
        'default_encoder_sem', 'default_decoder_sem', 'default_others_sem',
        'default_unet_dino'
    ])

    model.to("cuda")
    model.eval()


    dataset = TestDatasetv2(args, hr_folder=args.HR_test_folder, lr_folder=args.LR_test_folder)
    dataloader = torch.utils.data.DataLoader(
        dataset,
        batch_size=1,
        shuffle=False,
        num_workers=1
    )

    with torch.no_grad():
        for _, batch in tqdm(enumerate(dataloader), total=len(dataloader)):
            x_src = batch["lr"].cuda()
            x_tgt = batch["hr"].cuda()
            x_basename = batch["base_name"][0]

            assert x_src.shape[0] == 1, "Batch size must be 1"

            batch["prompt"] = [""]

            # Forward
            x_tgt_pred, *_ = model(x_src, x_tgt, batch=batch, args=args)
            #torch.save(x_tgt_pred, f"tensor_main_{x_basename.replace('.png', '.pt')}")

            # [-1,1] -> [0,1]
            x_tgt_pred = x_tgt_pred * 0.5 + 0.5
            x_src = x_src * 0.5 + 0.5
            x_tgt = x_tgt * 0.5 + 0.5

            # Remove batch dim
            x_tgt_pred = x_tgt_pred[0].cpu()
            x_src = x_src[0].cpu()
            x_tgt = x_tgt[0].cpu()

            # PIL
            output_pil = transforms.ToPILImage()(x_tgt_pred)
            input_pil = transforms.ToPILImage()(x_src)

            # Color alignment
            if args.align_method == "adain":
                output_pil = adain_color_fix(
                    target=output_pil,
                    source=input_pil
                )
            elif args.align_method == "wavelet":
                output_pil = wavelet_color_fix(
                    target=output_pil,
                    source=input_pil
                )

            # Save SR
            output_pil.save(os.path.join(pred_dir, x_basename))

            # Save GT
            #gt_pil = transforms.ToPILImage()(x_tgt)
            #gt_pil.save(os.path.join(gt_dir, x_basename))


if __name__ == "__main__":
    parser = argparse.ArgumentParser()

    parser.add_argument("--output_dir", type=str, default="./output")
    parser.add_argument("--lora_dir", type=str, default="./weights/checkpoints")
    parser.add_argument("--LR_test_folder", type=str, default="./LR")
    parser.add_argument("--HR_test_folder", type=str, default="/work/baraldimaticad/icpr_sr/osediff_dataset/RealSR/test_HR")


    parser.add_argument("--pretrained_model_path", type=str, default="/your/path/to/stable-diffusion-2-1-base")
    parser.add_argument("--seed", type=int, default=123)
    parser.add_argument("--process_size", type=int, default=512)
    parser.add_argument("--upscale", type=int, default=4)

    parser.add_argument("--neg_prompt", default="painting, oil painting, illustration, drawing, art, sketch, oil painting, cartoon, CG Style, 3D render, unreal engine, blurring, dirty, messy, worst quality, low quality, frames, watermark, signature, jpeg artifacts, deformed, lowres, over-smooth", type=str)

    parser.add_argument("--lora_rank_unet_pix", type=int, default=4)
    parser.add_argument("--lora_rank_unet_sem", type=int, default=4)
    parser.add_argument("--lora_rank_unet_dino", type=int, default=4)

    parser.add_argument("--timesteps1", type=float, default=1)
    parser.add_argument("--null_text_ratio", type=float, default=0.0)

    parser.add_argument("--align_method", type=str,
                        choices=["wavelet", "adain", "nofix"],
                        default="adain")

    args = parser.parse_args()

    main(args)