UniqAI/gaussian-splatting
1
0
Fork 0
mirror of https://github.com/graphdeco-inria/gaussian-splatting synced 2025-06-26 18:18:11 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

modified all errors

Browse Source
This commit is contained in:
sonbosung 2025-03-13 15:13:40 +09:00
parent 48ceb9419b
commit 2560a16632
6 changed files with 164 additions and 76 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

156
augment.py
View File

@ -1,5 +1,3 @@
import argparse
import numpy as np
from tqdm import tqdm
@ -13,6 +11,7 @@ from utils.aug_utils import *
def augment(colmap_path, image_path, augment_path, camera_order, visibility_aware_culling, compare_center_patch):
colmap_images, colmap_points3D, colmap_cameras = get_colmap_data(colmap_path)
np.seterr(divide='ignore', invalid='ignore')
sorted_keys = cluster_cameras(colmap_path, camera_order)
points3d = []
@ -29,11 +28,12 @@ def augment(colmap_path, image_path, augment_path, camera_order, visibility_awar
count = 0
roots = {}
pbar = tqdm(len(sorted_keys))
pbar = tqdm(range(len(sorted_keys)))
for view_idx in pbar:
view = sorted_keys[view_idx]
view_root, augmented_count = image_quadtree_augmentation(
view,
image_path,
colmap_cameras,
colmap_images,
colmap_points3D,
@ -42,25 +42,25 @@ def augment(colmap_path, image_path, augment_path, camera_order, visibility_awar
intrinsics_camera,
rotations_image,
translations_image,
visibility_aware_culling,
visibility_aware_culling=visibility_aware_culling,
)
count += augmented_count
pbar.set_description(f"{count} points augmented")
roots[view] = view_root
for view1_idx in tqdm(range(len(sorted_keys))):
for view2_idx in [view_idx + 6,
view_idx + 5,
view_idx + 4,
view_idx + 3,
view_idx + 2,
view_idx + 1,
view_idx - 1,
view_idx - 2,
view_idx - 3,
view_idx - 4,
view_idx - 5,
view_idx - 6]:
for view2_idx in [view1_idx + 6,
view1_idx + 5,
view1_idx + 4,
view1_idx + 3,
view1_idx + 2,
view1_idx + 1,
view1_idx - 1,
view1_idx - 2,
view1_idx - 3,
view1_idx - 4,
view1_idx - 5,
view1_idx - 6]:
if view2_idx > len(sorted_keys) - 1:
view2_idx = view2_idx - len(sorted_keys)
view1 = sorted_keys[view1_idx]
@ -93,6 +93,8 @@ def augment(colmap_path, image_path, augment_path, camera_order, visibility_awar
x, y = view1_sample_points_view2[i]
corresponding_node_type = None
error = None
# Case 1: Culling
if (view1_sample_points_view2_depth[i] < 0) | \
(view1_sample_points_view2[i, 0] < 0) | \
(view1_sample_points_view2[i, 0] >= image_view2.shape[1]) | \
@ -102,65 +104,85 @@ def augment(colmap_path, image_path, augment_path, camera_order, visibility_awar
corresponding_node_type = "culled"
matching_log.append([view2, corresponding_node_type, error])
continue
else:
view2_corresponding_node = find_leaf_node(view2_root, x, y)
if view2_corresponding_node is None:
corresponding_node_type = "missing"
else:
if view2_corresponding_node.unoccupied:
if view2_corresponding_node.depth_interpolated:
error = np.linalg.norm(view1_sample_points_view2_depth[i] - view2_corresponding_node.sampled_point_depth)
if error < 0.2 * view2_corresponding_node.sampled_point_depth:
if compare_center_patch:
try:
view1_sample_point_patch = image_view2[int(view1_sample_points_view2[i, 1])-1:\
int(view1_sample_points_view2[i,1])+2,
int(view1_sample_points_view2[i, 0])-1:\
int(view1_sample_points_view2[i,0])+2]
view2_corresponding_node_patch = image_view2[int(view2_corresponding_node.sampled_point_uv[1])-1:\
int(view2_corresponding_node.sampled_point_uv[1])+2,
int(view2_corresponding_node.sampled_point_uv[0])-1:\
int(view2_corresponding_node.sampled_point_uv[0])+2]
if compare_local_texture(view1_sample_point_patch, view2_corresponding_node_patch) > 0.5:
corresponding_node_type = "sampledrejected"
else:
corresponding_node_type = "sampled"
except IndexError:
corresponding_node_type = "sampledrejected"
# Case 2: Find corresponding node
view2_corresponding_node = find_leaf_node(view2_root, x, y)
if view2_corresponding_node is None:
corresponding_node_type = "missing"
matching_log.append([view2, corresponding_node_type, error])
continue
# Case 3: Process unoccupied node
if view2_corresponding_node.unoccupied:
if view2_corresponding_node.depth_interpolated:
error = np.linalg.norm(view1_sample_points_view2_depth[i] - view2_corresponding_node.sampled_point_depth)
if error < 0.2 * view2_corresponding_node.sampled_point_depth:
if compare_center_patch:
try:
view1_sample_point_patch = image_view2[int(view1_sample_points_view2[i, 1])-1:\
int(view1_sample_points_view2[i,1])+2,
int(view1_sample_points_view2[i, 0])-1:\
int(view1_sample_points_view2[i,0])+2]
view2_corresponding_node_patch = image_view2[int(view2_corresponding_node.sampled_point_uv[1])-1:\
int(view2_corresponding_node.sampled_point_uv[1])+2,
int(view2_corresponding_node.sampled_point_uv[0])-1:\
int(view2_corresponding_node.sampled_point_uv[0])+2]
if compare_local_texture(view1_sample_point_patch, view2_corresponding_node_patch) > 0.5:
corresponding_node_type = "sampledrejected"
else:
corresponding_node_type = "sampled"
else:
except IndexError:
corresponding_node_type = "sampledrejected"
else:
corresponding_node_type = "depthrejected"
corresponding_node_type = "sampled"
else:
corresponding_3d_depth = np.array(view2_corresponding_node.points3d_depths)
error = np.linalg.norm(view1_sample_points_view2_depth[i] - corresponding_3d_depth)
corresponding_node_type = "sampledrejected"
else:
corresponding_node_type = "depthrejected"
else:
corresponding_3d_depth = np.array(view2_corresponding_node.points3d_depths)
error = np.linalg.norm(view1_sample_points_view2_depth[i] - corresponding_3d_depth)
if np.min(error) < 0.2 * corresponding_3d_depth[np.argmin(error)]:
if compare_center_patch:
try:
point_3d_coord = points3d_view2_pixcoord[view2_corresponding_node.points3d_indices[np.argmin[error]]]
point_3d_patch = image_view2[int(point_3d_coord[1])-1:\
int(point_3d_coord[1])+2,
int(point_3d_coord[0])-1:\
int(point_3d_coord[0])+2]
view1_sample_point_patch = image_view2[int(view1_sample_points_view2[i, 1])-1:\
int(view1_sample_points_view2[i,1])+2,
int(view1_sample_points_view2[i, 0])-1:\
int(view1_sample_points_view2[i,0])+2]
if compare_local_texture(view1_sample_point_patch, point_3d_patch) > 0.5:
corresponding_node_type = "rejectedoccupied3d"
else:
corresponding_node_type = "occupied3d"
except IndexError:
corresponding_node_type = "rejectedoccupied3d"
if np.min(error) < 0.2 * corresponding_3d_depth[np.argmin(error)]:
if compare_center_patch:
try:
point_3d_coord = points3d_view2_pixcoord[view2_corresponding_node.points3d_indices[np.argmin(error)]]
point_3d_patch = image_view2[int(point_3d_coord[1])-1:\
int(point_3d_coord[1])+2,
int(point_3d_coord[0])-1:\
int(point_3d_coord[0])+2]
view1_sample_point_patch = image_view2[int(view1_sample_points_view2[i, 1])-1:\
int(view1_sample_points_view2[i,1])+2,
int(view1_sample_points_view2[i, 0])-1:\
int(view1_sample_points_view2[i,0])+2]
if compare_local_texture(view1_sample_point_patch, point_3d_patch) > 0.5:
corresponding_node_type = "rejectedoccupied3d"
else:
corresponding_node_type = "occupied3d"
else:
except IndexError:
corresponding_node_type = "rejectedoccupied3d"
matching_log.append([view2, corresponding_node_type, error])
else:
corresponding_node_type = "occupied3d"
else:
corresponding_node_type = "rejectedoccupied3d"
# 모든 경우에 대해 로그 추가
matching_log.append([view2, corresponding_node_type, error])
node_index = 0
view1_leaf_nodes = []
gather_leaf_nodes(view1_root, view1_leaf_nodes)
for node in view1_leaf_nodes:
if node.unoccupied:
if node.depth_interpolated:
node.matching_log[view2] = matching_log[node_index]
if matching_log[node_index][1] in ["depthrejected", "missing", "culled"]:
None
else:
node.inference_count += 1
node.rejection_count += 1 if matching_log[node_index][1] in ["rejectedoccupied3d",
"sampledrejected"] else 0
node_index += 1
sampled_points_total = []
sampled_points_rgb_total = []
@ -178,7 +200,7 @@ def augment(colmap_path, image_path, augment_path, camera_order, visibility_awar
sampled_points_total.append([node.sampled_point_world])
sampled_points_rgb_total.append([node.sampled_point_rgb])
sampled_points_uv_total.append([node.sampled_point_uv])
sampled_points_neighbors_uv_total.append([node.sampled_point_neighbors_uv])
sampled_points_neighbors_uv_total.append([node.sampled_point_neighbours_uv])
print("total_Sampled_points: ", len(sampled_points_total))
xyz = np.concatenate(sampled_points_total, axis=0)
rgb = np.concatenate(sampled_points_rgb_total, axis=0)

64
eval_mipnerf360.sh Normal file
View File

@ -0,0 +1,64 @@
for scene in bicycle flowers garden room stump
do
python augment.py --colmap_path /home/cvnar/disk4tb/360/${scene}/sparse/0 --image_path /home/cvnar/disk4tb/360/${scene}/images_4 \
--augment_path /home/cvnar/disk4tb/360_augmented/${scene}/sparse/0/points3D.bin \
--camera_order covisibility \
--visibility_aware_culling \
--compare_center_patch
python train.py -s /home/cvnar/disk4tb/360/${scene} -m ../experiments/360/${scene} \
-i images_4 \
--eval
python render.py -m ../experiments/360/${scene} --skip_train
python metrics.py -m ../experiments/360/${scene}
rm /home/cvnar/disk4tb/360_augmented/${scene}/sparse/0/points3D.ply
python train.py -s /home/cvnar/disk4tb/360_augmented/${scene} -m ../experiments/360_augmented/${scene} \
-i images_4 \
--eval \
--bundle_training \
--camera_order covisibility \
--enable_ds_lap \
--lambda_ds 1.2 \
--lambda_lap 0.4
python render.py -m ../experiments/360_augmented/${scene} --skip_train
python metrics.py -m ../experiments/360_augmented/${scene}
done
for scene in bonsai counter kitchen room
do
python augment.py --colmap_path /home/cvnar/disk4tb/360/${scene}/sparse/0 --image_path /home/cvnar/disk4tb/360/${scene}/images_2 \
--augment_path /home/cvnar/disk4tb/360_augmented/${scene}_augmented/sparse/0/points3D.bin \
--camera_order covisibility \
--visibility_aware_culling \
--compare_center_patch
python train.py -s /home/cvnar/disk4tb/360/${scene} -m ../experiments/360/${scene} \
-i images_2 \
--eval
python render.py -m ../experiments/360/${scene} --skip_train
python metrics.py -m ../experiments/360/${scene}
rm /home/cvnar/disk4tb/360_augmented/${scene}/sparse/0/points3D.ply
python train.py -s /home/cvnar/disk4tb/360_augmented/${scene} -m ../experiments/360_augmented/${scene} \
-i images_2 \
--eval \
--bundle_training \
--camera_order covisibility \
--enable_ds_lap \
--lambda_ds 1.2 \
--lambda_lap 0.4
python render.py -m ../experiments/360_augmented/${scene} --skip_train
python metrics.py -m ../experiments/360_augmented/${scene}
done

10
train.py
View File

@ -84,7 +84,7 @@ def training(dataset,
ema_Ll1depth_for_log = 0.0
if bundle_training:
sorted_keys = cluster_cameras(dataset.source_path, camera_order)
sorted_keys = cluster_cameras(os.path.join(dataset.source_path, 'sparse/0'), camera_order)
start_indices, cluster_sizes = bundle_start_index_generator(sorted_keys, 20)
n_interval = 0
@ -125,9 +125,9 @@ def training(dataset,
if not viewpoint_stack:
viewpoint_stack = scene.getTrainCameras().copy()
viewpoint_indices = list(range(len(viewpoint_stack)))
rand_idx = randint(0, len(viewpoint_indices) - 1)
viewpoint_cam = viewpoint_stack.pop(rand_idx)
vind = viewpoint_indices.pop(rand_idx)
rand_idx = randint(0, len(viewpoint_indices) - 1)
viewpoint_cam = viewpoint_stack.pop(rand_idx)
vind = viewpoint_indices.pop(rand_idx)
# Render
if (iteration - 1) == debug_from:
@ -364,8 +364,8 @@ if __name__ == "__main__":
args.checkpoint_iterations,
args.start_checkpoint,
args.debug_from,
args.bundle_training,
args.camera_order,
args.bundle_training,
args.enable_ds_lap,
args.lambda_ds,
args.lambda_lap)

0
utils/__init__.py Normal file
View File

2
utils/aug_utils.py
View File

@ -9,7 +9,7 @@ import rtree
from shapely.geometry import Point, box
from collections import defaultdict
from sklearn.decomposition import PCA
from colmap_utils import compute_extrinsics, get_colmap_data
from utils.colmap_utils import compute_extrinsics, get_colmap_data
from matplotlib import pyplot as plt
class Node:

8
utils/bundle_utils.py
View File

@ -3,7 +3,7 @@ import os
from colmap.scripts.python.read_write_model import *
import numpy as np
from collections import defaultdict
from colmap_utils import compute_extrinsics, compute_intrinsics, get_colmap_data
from utils.colmap_utils import compute_extrinsics, compute_intrinsics, get_colmap_data
import cv2
from sklearn.decomposition import PCA
@ -77,9 +77,9 @@ def create_sequence_from_covisibility_graph(covisibility_graph, min_covisibility
return sequence
def cluster_cameras(model_path, camera_order):
colmap_path = os.path.join(model_path, 'sparse/0')
def cluster_cameras(colmap_path, camera_order):
colmap_images, colmap_points3D, colmap_cameras = get_colmap_data(colmap_path)
print(camera_order)
if camera_order == 'covisibility':
covisibility_matrix, id_to_idx, idx_to_id = build_covisibility_matrix(colmap_images, colmap_points3D)
covisibility_graph = create_covisibility_graph(covisibility_matrix, idx_to_id)
@ -138,6 +138,8 @@ def cluster_cameras(model_path, camera_order):
sorted_cam_centers = cam_center_2d[sorted_indices]
sorted_keys = np.array(key)[sorted_indices]
print(sorted_keys)
return sorted_keys
def bundle_start_index_generator(sorted_keys, initial_interval):