Spaces:
Runtime error
Runtime error
jens
commited on
Commit
·
9917d3b
1
Parent(s):
95036e2
point-cloud
Browse files
app.py
CHANGED
@@ -2,14 +2,14 @@ import gradio as gr
|
|
2 |
from segment_anything import SamAutomaticMaskGenerator, sam_model_registry
|
3 |
import supervision as sv
|
4 |
from inference import DepthPredictor, SegmentPredictor
|
5 |
-
from utils import create_3d_obj
|
6 |
import numpy as np
|
7 |
|
8 |
|
9 |
def snap(image, video):
|
10 |
depth_predictor = DepthPredictor()
|
11 |
depth_result = depth_predictor.predict(image)
|
12 |
-
gltf_path =
|
13 |
#segment_predictor = SegmentPredictor()
|
14 |
#sam_result = segment_predictor.predict(image)
|
15 |
return [image, gltf_path, gltf_path]#[depth_result, gltf_path, gltf_path]
|
|
|
2 |
from segment_anything import SamAutomaticMaskGenerator, sam_model_registry
|
3 |
import supervision as sv
|
4 |
from inference import DepthPredictor, SegmentPredictor
|
5 |
+
from utils import create_3d_obj, create_3d_pc
|
6 |
import numpy as np
|
7 |
|
8 |
|
9 |
def snap(image, video):
|
10 |
depth_predictor = DepthPredictor()
|
11 |
depth_result = depth_predictor.predict(image)
|
12 |
+
gltf_path = create_3d_pc(np.array(image), depth_result)
|
13 |
#segment_predictor = SegmentPredictor()
|
14 |
#sam_result = segment_predictor.predict(image)
|
15 |
return [image, gltf_path, gltf_path]#[depth_result, gltf_path, gltf_path]
|
utils.py
CHANGED
@@ -50,4 +50,38 @@ def create_3d_obj(rgb_image, depth_image, depth=10):
|
|
50 |
gltf_path = './image.gltf'
|
51 |
o3d.io.write_triangle_mesh(
|
52 |
gltf_path, mesh_crop, write_triangle_uvs=True)
|
53 |
-
return gltf_path
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
gltf_path = './image.gltf'
|
51 |
o3d.io.write_triangle_mesh(
|
52 |
gltf_path, mesh_crop, write_triangle_uvs=True)
|
53 |
+
return gltf_path
|
54 |
+
|
55 |
+
|
56 |
+
def create_3d_pc(rgb_image, depth_image, image_path, depth=10):
|
57 |
+
depth_image = depth_image.astype(np.float32) # Convert depth map to float32
|
58 |
+
depth_o3d = o3d.geometry.Image(depth_image)
|
59 |
+
image_o3d = o3d.geometry.Image(rgb_image)
|
60 |
+
rgbd_image = o3d.geometry.RGBDImage.create_from_color_and_depth(
|
61 |
+
image_o3d, depth_o3d, convert_rgb_to_intensity=False)
|
62 |
+
|
63 |
+
w = int(depth_image.shape[1])
|
64 |
+
h = int(depth_image.shape[0])
|
65 |
+
|
66 |
+
# Specify camera intrinsic parameters (modify based on actual camera)
|
67 |
+
fx = 500
|
68 |
+
fy = 500
|
69 |
+
cx = w / 2
|
70 |
+
cy = h / 2
|
71 |
+
|
72 |
+
camera_intrinsic = o3d.camera.PinholeCameraIntrinsic(w, h, fx, fy, cx, cy)
|
73 |
+
|
74 |
+
pcd = o3d.geometry.PointCloud.create_from_rgbd_image(
|
75 |
+
rgbd_image, camera_intrinsic)
|
76 |
+
|
77 |
+
print('Estimating normals...')
|
78 |
+
pcd.estimate_normals(
|
79 |
+
search_param=o3d.geometry.KDTreeSearchParamHybrid(radius=0.01, max_nn=30))
|
80 |
+
pcd.orient_normals_towards_camera_location(
|
81 |
+
camera_location=np.array([0., 0., 1000.]))
|
82 |
+
|
83 |
+
# Save the point cloud as a PLY file
|
84 |
+
ply_path = image_path.with_suffix('.ply')
|
85 |
+
o3d.io.write_point_cloud(str(ply_path), pcd)
|
86 |
+
|
87 |
+
return str(ply_path) # Return the file path where the PLY file is saved
|