Spaces:
Running
on
Zero
Running
on
Zero
File size: 5,470 Bytes
184193d 5bb9a7d 184193d c070b3c 184193d c070b3c 184193d c070b3c 184193d |
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 |
import os
import importlib
import imageio
import torch
import rembg
import numpy as np
import PIL.Image
from PIL import Image
from typing import Any
from torchvision import transforms
def instantiate_from_config(config):
if not "target" in config:
if config == '__is_first_stage__':
return None
elif config == "__is_unconditional__":
return None
raise KeyError("Expected key `target` to instantiate.")
return get_obj_from_str(config["target"])(**config.get("params", dict()))
def get_obj_from_str(string, reload=False):
module, cls = string.rsplit(".", 1)
if reload:
module_imp = importlib.import_module(module)
importlib.reload(module_imp)
return getattr(importlib.import_module(module, package=None), cls)
# def resize_without_crop(pil_image, target_width, target_height):
# resized_image = pil_image.resize((target_width, target_height), Image.LANCZOS)
# return np.array(resized_image)[:, :, :3]
# @torch.inference_mode()
# def numpy2pytorch(imgs):
# h = torch.from_numpy(np.stack(imgs, axis=0)).float() / 255.0 * 2.0 - 1.0
# h = h.movedim(-1, 1)
# return h
# @torch.inference_mode()
# def remove_background(
# image: PIL.Image.Image,
# rembg: Any = None,
# force: bool = False,
# **rembg_kwargs,
# ) -> PIL.Image.Image:
# do_remove = True
# if image.mode == "RGBA" and image.getextrema()[3][0] < 255:
# do_remove = False
# do_remove = do_remove or force
# if do_remove:
# W, H = image.size
# k = (256.0 / float(H * W)) ** 0.5
# feed = resize_without_crop(image, int(64 * round(W * k)), int(64 * round(H * k)))
# feed = numpy2pytorch([feed]).to(device=rembg.device, dtype=torch.float32)
# alpha = rembg(feed)[0][0]
# alpha = torch.nn.functional.interpolate(alpha, size=(H, W), mode="bilinear")
# alpha = alpha.squeeze().clamp(0, 1)
# alpha = (alpha * 255).cpu().data.numpy().astype(np.uint8)
# alpha = Image.fromarray(alpha)
# no_bg_image = Image.new("RGBA", alpha.size, (0, 0, 0, 0))
# no_bg_image.paste(image, mask=alpha)
# image = no_bg_image
# return image
@torch.inference_mode()
def remove_background(
image: PIL.Image.Image,
rembg: Any = None,
force: bool = False,
**rembg_kwargs,
) -> PIL.Image.Image:
do_remove = True
if image.mode == "RGBA" and image.getextrema()[3][0] < 255:
do_remove = False
do_remove = do_remove or force
if do_remove:
transform_image = transforms.Compose([
transforms.Resize((1024, 1024)),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])
image = image.convert('RGB')
input_images = transform_image(image).unsqueeze(0).to(rembg.device)
with torch.no_grad():
preds = rembg(input_images)[-1].sigmoid().cpu()
pred = preds[0].squeeze()
pred_pil = transforms.ToPILImage()(pred)
mask = pred_pil.resize(image.size)
image.putalpha(mask)
return image
# def remove_background(image: PIL.Image.Image,
# rembg_session: Any = None,
# force: bool = False,
# **rembg_kwargs,
# ) -> PIL.Image.Image:
# do_remove = True
# if image.mode == "RGBA" and image.getextrema()[3][0] < 255:
# do_remove = False
# do_remove = do_remove or force
# if do_remove:
# image = rembg.remove(image, session=rembg_session, **rembg_kwargs)
# return image
def resize_foreground(
image: PIL.Image.Image,
ratio: float,
) -> PIL.Image.Image:
image = np.array(image)
assert image.shape[-1] == 4
alpha = np.where(image[..., 3] > 0)
y1, y2, x1, x2 = (
alpha[0].min(),
alpha[0].max(),
alpha[1].min(),
alpha[1].max(),
)
# crop the foreground
fg = image[y1:y2, x1:x2]
# pad to square
size = max(fg.shape[0], fg.shape[1])
ph0, pw0 = (size - fg.shape[0]) // 2, (size - fg.shape[1]) // 2
ph1, pw1 = size - fg.shape[0] - ph0, size - fg.shape[1] - pw0
new_image = np.pad(
fg,
((ph0, ph1), (pw0, pw1), (0, 0)),
mode="constant",
constant_values=((0, 0), (0, 0), (0, 0)),
)
# compute padding according to the ratio
new_size = int(new_image.shape[0] / ratio)
# pad to size, double side
ph0, pw0 = (new_size - size) // 2, (new_size - size) // 2
ph1, pw1 = new_size - size - ph0, new_size - size - pw0
new_image = np.pad(
new_image,
((ph0, ph1), (pw0, pw1), (0, 0)),
mode="constant",
constant_values=((0, 0), (0, 0), (0, 0)),
)
new_image = Image.fromarray(new_image)
return new_image
def rgba_to_white_background(image: PIL.Image.Image) -> torch.Tensor:
image = np.asarray(image, dtype=np.float32) / 255.0
image = torch.from_numpy(image).movedim(2, 0).float()
image, alpha = image.split([3, 1], dim=0)
image = image * alpha + torch.ones_like(image) * (1 - alpha)
return image, alpha
def save_video(
frames: torch.Tensor,
output_path: str,
fps: int = 30,
) -> None:
# images: (N, C, H, W)
frames = [(frame.permute(1, 2, 0).cpu().numpy() * 255).astype(np.uint8) for frame in frames]
writer = imageio.get_writer(output_path, mode='I', fps=fps, codec='libx264')
for frame in frames:
writer.append_data(frame)
writer.close() |