Spaces:
Sleeping
Sleeping
File size: 2,675 Bytes
42b0b31 |
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 |
#!/usr/bin/env python
# coding: utf-8
# In[37]:
import torch
import torch.nn as nn
from functools import partial
#import clip
from einops import rearrange, repeat
from glob import glob
from PIL import Image
from torchvision import transforms as T
from tqdm import tqdm
import pickle
import numpy as np
import os
from transformers import AutoProcessor, CLIPVisionModelWithProjection, CLIPProcessor, CLIPModel
device = 'cuda:0'
#model = CLIPVisionModelWithProjection.from_pretrained("openai/clip-vit-large-patch14").to(device)
#processor = CLIPProcessor.from_pretrained("openai/clip-vit-large-patch14")
class ClipImageEncoder(nn.Module):
def __init__(self):
super().__init__()
self.emb_dim = (1, 257, 1024)
self.model = CLIPVisionModelWithProjection.from_pretrained("openai/clip-vit-large-patch14")
self.processor = CLIPProcessor.from_pretrained("openai/clip-vit-large-patch14")
self.model = self.model.eval()
for param in self.parameters():
param.requires_grad = False
@torch.no_grad()
def forward(self, x):
ret = self.model(x)
return ret.last_hidden_state, ret.image_embeds
def preprocess(self, style_image):
# if os.path.exists(style_file):
# style_image = Image.open(style_file)
# else:
# style_image = Image.fromarray(np.zeros((224,224,3), dtype=np.uint8))
x = torch.tensor(np.array(self.processor.image_processor(style_image).pixel_values))
return x
def postprocess(self, x): # return numpy
return x.detach().cpu().squeeze(0).numpy()
if __name__ == '__main__':
device = 'cuda:1'
style_files = glob("/home/soon/datasets/deepfashion_inshop/styles_default/**/*.jpg", recursive=True)
style_files = [x for x in style_files if x.split('/')[-1]!='background.jpg']
clip_model = ClipImageEncoder().to(device)
for style_file in tqdm(style_files[24525:]):
style_image = Image.open(style_file)
emb_local, emb_global = clip_model(clip_model.preprocess(style_image).to(device))
emb_local = clip_model.postprocess(emb_local)
emb_global = clip_model.postprocess(emb_global)
#x = torch.tensor(np.array(processor.image_processor(style_image).pixel_values))
#emb = model(x.to(device)).last_hidden_state
#emb = emb.detach().cpu().squeeze(0).numpy()
emb_file = style_file.replace('.jpg','_hidden.p')
with open(emb_file, 'wb') as file:
pickle.dump(emb_local, file)
emb_file = style_file.replace('.jpg','.p')
with open(emb_file, 'wb') as file:
pickle.dump(emb_global, file)
|