code/model_vis_tools/pca_dinov2.py

import os
import torch
import numpy as np
from PIL import Image
from sklearn.decomposition import PCA
from math import sqrt
from torchvision import transforms
import matplotlib.pyplot as plt
from open_clip.transform import ResizeLongest, _convert_to_rgb, ToTensor, Normalize
from torchvision.transforms import InterpolationMode
# Ensure reproducibility
torch.manual_seed(42)
np.random.seed(42)

# Define constants
mean = [0.485, 0.456, 0.406]
std = [0.229, 0.224, 0.225]
target_size = (560, 560)

# Define transforms
normalize = Normalize(mean=mean, std=std)
DINO_transform = transforms.Compose([
    transforms.Resize((560,560),interpolation=InterpolationMode.BILINEAR),
    _convert_to_rgb,
    ToTensor(),
    normalize
])

# PCA visualization function
def plot_pca(f, path, target_size):
    """
    Visualize PCA for the given features.

    Args:
        f (numpy.ndarray): Feature array of shape [N, D].
        path (str): Path to save the PCA visualization.
        target_size (tuple): Target size of the output image.
    """
    pca = PCA(n_components=3)
    pca.fit(f)
    pca_img = pca.transform(f)  # n x 3
    h = w = int(sqrt(pca_img.shape[0]))
    pca_img = pca_img.reshape(h, w, 3)
    pca_img_min = pca_img.min(axis=(0, 1))
    pca_img_max = pca_img.max(axis=(0, 1))
    pca_img = (pca_img - pca_img_min) / (pca_img_max - pca_img_min + 1e-8)  # Normalize
    pca_img = Image.fromarray((pca_img * 255).astype(np.uint8))
    pca_img = transforms.Resize(target_size, interpolation=transforms.InterpolationMode.NEAREST)(pca_img)
    pca_img.save(path)

# DINOv2 model loading function
def build_DINOv2():
    model_name = 'dinov2_vitb14_reg'
    hub_path = '/mnt/SSD8T/home/wjj/.cache/torch/hub/facebookresearch_dinov2_main'
    try:
        vfm = torch.hub.load(hub_path, model_name, source='local').half()
    except Exception as e:
        raise RuntimeError(f"Failed to load DINOv2 model '{model_name}': {e}")
    return vfm

# Main function for DINOv2 PCA visualization
def visualize_dinov2_pca(image_path, output_dir):
    """
    Visualize PCA for DINOv2 features.

    Args:
        image_path (str): Path to the input image.
        output_dir (str): Directory to save the PCA visualization.
    """
    # Ensure output directory exists
    if not os.path.exists(output_dir):
        os.makedirs(output_dir)

    # Load DINOv2 model
    device = "cuda" if torch.cuda.is_available() else "cpu"
    dino = build_DINOv2().to(device)

    # Load and preprocess image
    image = Image.open(image_path)
    image_tensor = DINO_transform(image).unsqueeze(0).to(torch.float16).to(device)

    # Extract DINOv2 features
    with torch.no_grad():
        dino_feats_raw = dino.get_intermediate_layers(image_tensor, reshape=True)[0]
        dino_feats_raw = dino_feats_raw.flatten(start_dim=-2).transpose(-2, -1)  # [1, N, D]

    # Prepare features for PCA
    dino_feats_np = dino_feats_raw[0].cpu().numpy()

    # Save PCA visualization
    pca_path = os.path.join(output_dir, "dinov2_pca.png")
    plot_pca(dino_feats_np, pca_path, target_size)

    print(f"PCA visualization saved at: {pca_path}")

# Example usage
if __name__ == "__main__":
    image_path = "demo_images/bird4.jpg"  # Replace with your image path
    output_dir = "dinov2_vis"
    visualize_dinov2_pca(image_path, output_dir)