File size: 5,765 Bytes
01fdb75 | 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 | """
Compare different sampling strategies: Euler-50, Euler-100, Euler-200, Heun-50
Layout: Mask | Euler-50 | Euler-100 | Euler-200 | Heun-50 | Real
"""
import sys
sys.path.insert(0, "/data/sichengli/Code/PixelGen")
import torch
import numpy as np
from PIL import Image, ImageDraw, ImageFont
import torchvision.transforms as transforms
import torchvision.transforms.functional as TF
import os, random
from src.models.transformer.JiT_medical import JiTMedical
from src.diffusion.flow_matching.sampling_medical import EulerSamplerMedical, HeunSamplerMedical
from src.diffusion.flow_matching.scheduling import LinearScheduler
from src.diffusion.base.guidance import simple_guidance_fn
device = torch.device("cuda:0")
# Load model
print("Loading model...")
ckpt_path = "/data/sichengli/Code/PixelGen/medical_workdirs/exp_PixelGen_Medical_B16/epoch=236-step=100000.ckpt"
ckpt = torch.load(ckpt_path, map_location="cpu", weights_only=False)
state_dict = ckpt["state_dict"]
model = JiTMedical(
input_size=256, patch_size=16, in_channels=3,
hidden_size=768, depth=12, num_heads=12, mlp_ratio=4.0,
attn_drop=0.0, proj_drop=0.1, num_classes=1,
use_bottleneck=True, bottleneck_dim=128,
in_context_len=32, in_context_start=4, mask_in_channels=1
)
ema_state = {}
for k, v in state_dict.items():
if k.startswith("ema_denoiser."):
new_k = k.replace("ema_denoiser.", "").replace("_orig_mod.", "")
ema_state[new_k] = v
model.load_state_dict(ema_state, strict=False)
model = model.to(device).eval().to(torch.float32)
# Shared sampler kwargs
sampler_kwargs = dict(
guidance=2.0, timeshift=1.0,
guidance_interval_min=0.1, guidance_interval_max=0.9,
scheduler=LinearScheduler(), w_scheduler=LinearScheduler(),
guidance_fn=simple_guidance_fn,
)
# Build samplers
samplers = {
"Euler-50": EulerSamplerMedical(num_steps=50, **sampler_kwargs).to(device),
"Euler-100": EulerSamplerMedical(num_steps=100, **sampler_kwargs).to(device),
"Euler-200": EulerSamplerMedical(num_steps=200, **sampler_kwargs).to(device),
"Heun-50": HeunSamplerMedical(num_steps=50, **sampler_kwargs).to(device),
}
# Load 6 samples
data_root = "/data2/sichengli/Data/test/Segmentation/OCTA500"
img_dir = os.path.join(data_root, "images")
mask_dir = os.path.join(data_root, "masks")
all_files = sorted([f for f in os.listdir(img_dir) if f.endswith(".png") and not f.startswith("thumb")])
random.seed(456)
selected = random.sample(all_files, 6)
images_list, masks_list = [], []
for fname in selected:
img = Image.open(os.path.join(img_dir, fname)).convert("L")
img = TF.resize(img, (256, 256))
images_list.append(TF.to_tensor(img).repeat(3, 1, 1))
mask = Image.open(os.path.join(mask_dir, fname)).convert("L")
mask = TF.resize(mask, (256, 256), interpolation=transforms.InterpolationMode.NEAREST)
masks_list.append(TF.to_tensor(mask))
real_images = torch.stack(images_list)
masks_tensor = torch.stack(masks_list)
# Use same noise for fair comparison
torch.manual_seed(42)
shared_noise = torch.randn(6, 3, 256, 256, device=device)
# Generate with each sampler
results = {}
for name, sampler in samplers.items():
print(f"Generating with {name}...")
with torch.no_grad():
noise = shared_noise.clone()
x_trajs, _ = sampler._impl_sampling(model, noise, None, None, mask=masks_tensor.to(device))
gen = x_trajs[-1].clamp(-1, 1) * 0.5 + 0.5
results[name] = gen.cpu()
print(f" Done. range=[{gen.min():.3f}, {gen.max():.3f}]")
# Create comparison grid
# Columns: Mask | Euler-50 | Euler-100 | Euler-200 | Heun-50 | Real
col_labels = ["Mask", "Euler-50", "Euler-100", "Euler-200", "Heun-50", "Real"]
n_rows = 6
n_cols = len(col_labels)
h, w = 256, 256
pad = 4
label_h = 36
canvas_w = n_cols * w + (n_cols + 1) * pad
canvas_h = n_rows * h + (n_rows + 1) * pad + label_h
canvas = np.ones((canvas_h, canvas_w, 3), dtype=np.uint8) * 30
# Column labels
try:
font = ImageFont.truetype("/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf", 20)
except Exception:
font = ImageFont.load_default()
pil_canvas = Image.fromarray(canvas)
draw = ImageDraw.Draw(pil_canvas)
for col, label in enumerate(col_labels):
x_pos = pad + col * (w + pad) + w // 2
bbox = draw.textbbox((0, 0), label, font=font)
text_w = bbox[2] - bbox[0]
draw.text((x_pos - text_w // 2, 8), label, fill=(255, 255, 255), font=font)
canvas = np.array(pil_canvas)
# Colormap for mask
color_map = {
0: (0, 0, 0),
50: (255, 80, 80),
100: (80, 255, 80),
150: (80, 80, 255),
200: (255, 255, 80),
250: (255, 80, 255),
}
for row in range(n_rows):
y = label_h + pad + row * (h + pad)
for col_idx, col_name in enumerate(col_labels):
x = pad + col_idx * (w + pad)
if col_name == "Mask":
m = masks_tensor[row, 0].numpy()
m_uint8 = (m * 255).astype(np.uint8)
m_colored = np.zeros((h, w, 3), dtype=np.uint8)
for val, color in color_map.items():
mask_region = np.abs(m_uint8.astype(int) - val) < 13
m_colored[mask_region] = color
canvas[y:y+h, x:x+w] = m_colored
elif col_name == "Real":
r = real_images[row].permute(1, 2, 0).numpy()
r = (r * 255).clip(0, 255).astype(np.uint8)
canvas[y:y+h, x:x+w] = r
else:
g = results[col_name][row].permute(1, 2, 0).numpy()
g = (g * 255).clip(0, 255).astype(np.uint8)
canvas[y:y+h, x:x+w] = g
out_dir = "/data/sichengli/Code/PixelGen/medical_workdirs/exp_PixelGen_Medical_B16/val_samples"
out_path = os.path.join(out_dir, "sampling_compare.png")
Image.fromarray(canvas).save(out_path)
print(f"\nSaved: {out_path} ({canvas_w}x{canvas_h})")
|