File size: 9,641 Bytes
fb6c2da
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
import argparse
import logging
import os
import os.path as osp
import time

import cv2
import matplotlib.pyplot as plt
import numpy as np
import torch
import torch.nn as nn
from basicsr.utils import (get_env_info, get_root_logger, get_time_str,
                           img2tensor, scandir, tensor2img)
from basicsr.utils.options import copy_opt_file, dict2str
from omegaconf import OmegaConf
from PIL import Image
from pytorch_lightning import seed_everything

from dataset_coco import dataset_coco, dataset_coco_mask_color_sig
from dist_util import get_bare_model, init_dist, master_only
from ldm.models.diffusion.ddim import DDIMSampler
from ldm.models.diffusion.dpm_solver import DPMSolverSampler
from ldm.models.diffusion.plms import PLMSSampler
from ldm.modules.encoders.adapter import Adapter
from ldm.util import instantiate_from_config


def load_model_from_config(config, ckpt, verbose=False):
    print(f"Loading model from {ckpt}")
    pl_sd = torch.load(ckpt, map_location="cpu")
    if "global_step" in pl_sd:
        print(f"Global Step: {pl_sd['global_step']}")
    sd = pl_sd["state_dict"]
    model = instantiate_from_config(config.model)
    m, u = model.load_state_dict(sd, strict=False)
    if len(m) > 0 and verbose:
        print("missing keys:")
        print(m)
    if len(u) > 0 and verbose:
        print("unexpected keys:")
        print(u)

    model.cuda()
    model.eval()
    return model

@master_only
def mkdir_and_rename(path):
    """mkdirs. If path exists, rename it with timestamp and create a new one.
    Args:
        path (str): Folder path.
    """
    if osp.exists(path):
        new_name = path + '_archived_' + get_time_str()
        print(f'Path already exists. Rename it to {new_name}', flush=True)
        os.rename(path, new_name)
    os.makedirs(path, exist_ok=True)
    os.makedirs(osp.join(experiments_root, 'models'))
    os.makedirs(osp.join(experiments_root, 'training_states'))
    os.makedirs(osp.join(experiments_root, 'visualization'))

def load_resume_state(opt):
    resume_state_path = None
    if opt.auto_resume:
        state_path = osp.join('experiments', opt.name, 'training_states')
        if osp.isdir(state_path):
            states = list(scandir(state_path, suffix='state', recursive=False, full_path=False))
            if len(states) != 0:
                states = [float(v.split('.state')[0]) for v in states]
                resume_state_path = osp.join(state_path, f'{max(states):.0f}.state')
                opt.resume_state_path = resume_state_path

    if resume_state_path is None:
        resume_state = None
    else:
        device_id = torch.cuda.current_device()
        resume_state = torch.load(resume_state_path, map_location=lambda storage, loc: storage.cuda(device_id))

    return resume_state

parser = argparse.ArgumentParser()
parser.add_argument(
        "--prompt",
        type=str,
        nargs="?",
        default="A black Honda motorcycle parked in front of a garage"
)
parser.add_argument(
        "--neg_prompt",
        type=str,
        default="ugly, tiling, poorly drawn hands, poorly drawn feet, poorly drawn face, out of frame, extra limbs, disfigured, deformed, body out of frame, bad anatomy, watermark, signature, cut off, low contrast, underexposed, overexposed, bad art, beginner, amateur, distorted face"
)
parser.add_argument(
        "--path_cond",
        type=str,
        default="examples/seg/motor.png"
)
parser.add_argument(
    "--bsize",
    type=int,
    default=8,
    help="the prompt to render"
)
parser.add_argument(
    "--epochs",
    type=int,
    default=10000,
    help="the prompt to render"
)
parser.add_argument(
    "--device",
    type=str,
    default="cuda"
)
parser.add_argument(
    "--num_workers",
    type=int,
    default=8,
    help="the prompt to render"
)
parser.add_argument(
    "--use_shuffle",
    type=bool,
    default=True,
    help="the prompt to render"
)
parser.add_argument(
        "--dpm_solver",
        action='store_true',
        help="use dpm_solver sampling",
)
parser.add_argument(
        "--plms",
        action='store_true',
        help="use plms sampling",
)
parser.add_argument(
        "--auto_resume",
        action='store_true',
        help="use plms sampling",
)
parser.add_argument(
        "--ckpt",
        type=str,
        default="models/sd-v1-4.ckpt",
        help="path to checkpoint of model",
)
parser.add_argument(
        "--ckpt_ad",
        type=str,
        default="models/t2iadapter_seg_sd14v1.pth"
)
parser.add_argument(
        "--config",
        type=str,
        default="configs/stable-diffusion/test_mask.yaml",
        help="path to config which constructs model",
)
parser.add_argument(
        "--print_fq",
        type=int,
        default=100,
        help="path to config which constructs model",
)
parser.add_argument(
        "--H",
        type=int,
        default=512,
        help="image height, in pixel space",
)
parser.add_argument(
    "--W",
    type=int,
    default=512,
    help="image width, in pixel space",
)
parser.add_argument(
    "--C",
    type=int,
    default=4,
    help="latent channels",
)
parser.add_argument(
    "--f",
    type=int,
    default=8,
    help="downsampling factor",
)
parser.add_argument(
        "--ddim_steps",
        type=int,
        default=50,
        help="number of ddim sampling steps",
)
parser.add_argument(
        "--n_samples",
        type=int,
        default=10,
        help="how many samples to produce for each given prompt. A.k.a. batch size",
)
parser.add_argument(
        "--ddim_eta",
        type=float,
        default=0.0,
        help="ddim eta (eta=0.0 corresponds to deterministic sampling",
)
parser.add_argument(
        "--scale",
        type=float,
        default=7.5,
        help="unconditional guidance scale: eps = eps(x, empty) + scale * (eps(x, cond) - eps(x, empty))",
)
parser.add_argument(
        "--gpus",
        default=[0,1,2,3],
        help="gpu idx",
)
parser.add_argument(
        '--local_rank',
        default=-1,
        type=int,
        help='node rank for distributed training'
)
parser.add_argument(
        '--launcher',
        default='pytorch',
        type=str,
        help='node rank for distributed training'
)
opt = parser.parse_args()

if __name__ == '__main__':
    # seed_everything(42)
    config = OmegaConf.load(f"{opt.config}")
    opt.name = config['name']
    device=opt.device

    # stable diffusion
    model = load_model_from_config(config, f"{opt.ckpt}").to(device)

    # Adaptor
    model_ad = Adapter(cin=int(3*64), channels=[320, 640, 1280, 1280][:4], nums_rb=2, ksize=1, sk=True, use_conv=False).to(device)
    model_ad.load_state_dict(torch.load(opt.ckpt_ad))

    experiments_root = osp.join('experiments', opt.name)

    # resume state
    resume_state = load_resume_state(opt)
    if resume_state is None:
        mkdir_and_rename(experiments_root)

    # copy the yml file to the experiment root
    copy_opt_file(opt.config, experiments_root)

    # WARNING: should not use get_root_logger in the above codes, including the called functions
    # Otherwise the logger will not be properly initialized
    log_file = osp.join(experiments_root, f"train_{opt.name}_{get_time_str()}.log")
    logger = get_root_logger(logger_name='basicsr', log_level=logging.INFO, log_file=log_file)
    logger.info(get_env_info())
    logger.info(dict2str(config))

    for v_idx in range(opt.n_samples):
        with torch.no_grad():
            if opt.dpm_solver:
                sampler = DPMSolverSampler(model)
            elif opt.plms:
                sampler = PLMSSampler(model)
            else:
                sampler = DDIMSampler(model)
            c = model.get_learned_conditioning([opt.prompt])

            # costumer input
            mask = cv2.imread(opt.path_cond)
            mask = cv2.resize(mask,(512,512))
            mask = img2tensor(mask, bgr2rgb=True, float32=True)/255.
            mask = mask.unsqueeze(0)

            im_mask = tensor2img(mask)
            cv2.imwrite(os.path.join(experiments_root, 'visualization', 'mask_idx%04d.png'%(v_idx)), im_mask)

            features_adapter = model_ad(mask.to(device))

            shape = [opt.C, opt.H // opt.f, opt.W // opt.f]

            samples_ddim, intermediates = sampler.sample(S=opt.ddim_steps,
                                                conditioning=c,
                                                batch_size=1,
                                                shape=shape,
                                                verbose=False,
                                                unconditional_guidance_scale=opt.scale,
                                                unconditional_conditioning=model.get_learned_conditioning([opt.neg_prompt]),
                                                eta=opt.ddim_eta,
                                                x_T=None,
                                                features_adapter1=features_adapter,
                                                mode = 'mask'
                                                )

            x_samples_ddim = model.decode_first_stage(samples_ddim)
            x_samples_ddim = torch.clamp((x_samples_ddim + 1.0) / 2.0, min=0.0, max=1.0)
            x_samples_ddim = x_samples_ddim.cpu().permute(0, 2, 3, 1).numpy()
            for id_sample, x_sample in enumerate(x_samples_ddim):
                x_sample = 255.*x_sample
                img = x_sample.astype(np.uint8)
                cv2.imwrite(os.path.join(experiments_root, 'visualization', 'sample_idx%04d_s%04d.png'%(v_idx, id_sample)), img[:,:,::-1])