File size: 2,471 Bytes
ed78524
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import streamlit as st
import numpy as np
import torch
import torchvision.transforms as T
from PIL import Image
import matplotlib.pyplot as plt
import open3d as o3d

# Load MiDaS model
@st.cache_resource
def load_model():
    model_type = "DPT_Large"  # Use DPT_Large for higher accuracy
    model = torch.hub.load("intel-isl/MiDaS", model_type)
    model.eval()
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    model.to(device)

    midas_transforms = torch.hub.load("intel-isl/MiDaS", "transforms")
    transform = midas_transforms.default_transform
    return model, transform, device

model, transform, device = load_model()

# Streamlit app UI
st.title("2D to 3D Image Converter")
st.write("Upload a 2D image to generate its 3D depth map and point cloud.")

# File uploader
uploaded_file = st.file_uploader("Upload an Image", type=["jpg", "jpeg", "png"])
if uploaded_file:
    # Load the image
    input_image = Image.open(uploaded_file).convert("RGB")
    st.image(input_image, caption="Uploaded Image", use_column_width=True)

    # Preprocess the image
    input_batch = transform(input_image).unsqueeze(0).to(device)

    # Predict depth map
    st.write("Generating Depth Map...")
    with torch.no_grad():
        prediction = model(input_batch)
        depth_map = torch.nn.functional.interpolate(
            prediction.unsqueeze(1),
            size=input_image.size[::-1],
            mode="bicubic",
            align_corners=False,
        ).squeeze().cpu().numpy()

    # Display depth map
    st.write("Depth Map:")
    fig, ax = plt.subplots()
    ax.imshow(depth_map, cmap="plasma")
    ax.axis("off")
    st.pyplot(fig)

    # Generate 3D point cloud
    st.write("Generating 3D Point Cloud...")
    h, w = depth_map.shape
    xx, yy = np.meshgrid(np.arange(w), np.arange(h))
    points = np.stack((xx, yy, depth_map), axis=-1).reshape(-1, 3)

    # Normalize points
    points -= points.mean(axis=0)
    points /= points.max(axis=0)

    # Create Open3D point cloud
    pcd = o3d.geometry.PointCloud()
    pcd.points = o3d.utility.Vector3dVector(points)

    # Save point cloud for download
    output_file = "point_cloud.ply"
    o3d.io.write_point_cloud(output_file, pcd)
    st.write("3D Point Cloud Generated!")
    st.download_button(
        label="Download 3D Point Cloud (.ply)",
        data=open(output_file, "rb").read(),
        file_name="3d_point_cloud.ply",
        mime="application/octet-stream",
    )