Views
No views yet
ColorJitter). تظهر الصور تبايناً عالياً، وحوافاً حادة، وانعكاسات ضوئية ممتازة دون حدوث حفظ ظاهري (Overfitting).
1import torch
2from diffusers import DDPMPipeline
3from PIL import Image
4
5device = "cuda" if torch.cuda.is_available() else "cpu"
6model_path = r"C:\Path\To\Your\Model_Folder"
7
8pipeline = DDPMPipeline.from_pretrained(model_path).to(device)
9images = pipeline(batch_size=64, num_inference_steps=75).images
10
11canvas = Image.new('RGB', (512, 512))
12idx = 0
13for r in range(8):
14 for c in range(8):
15 canvas.paste(images[idx], (c * 64, r * 64))
16 idx += 1
17
18canvas.save("output_windows.png")
19canvas.show()1import os
2import shutil
3import torch
4from diffusers import DDPMPipeline
5from PIL import Image
6from google.colab import files
7
8device = "cuda" if torch.cuda.is_available() else "cpu"
9model_path = "/content/AI_Fish_Model"
10
11if not os.path.exists(model_path) or not os.listdir(model_path):
12 os.makedirs(model_path, exist_ok=True)
13 uploaded = files.upload()
14 for filename in uploaded.keys():
15 shutil.move(filename, os.path.join(model_path, filename))
16
17pipeline = DDPMPipeline.from_pretrained(model_path).to(device)
18images = pipeline(batch_size=64, num_inference_steps=75).images
19
20canvas = Image.new('RGB', (512, 512))
21idx = 0
22for r in range(8):
23 for c in range(8):
24 canvas.paste(images[idx], (c * 64, r * 64))
25 idx += 1
26
27canvas.save("output_colab.png")
28canvas.show()