hofee/nbv_reconstruction
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Compare commits

base: hofee:60c935749146471c3a5dc0c4fd9523b0ddc63843
Branches Tags
hofee:master
hofee:ab_new_partial_global
hofee:new_partial_global
hofee:ab_global_only
hofee:hemisphere_random
hofee:ab_global_and_partial_global
hofee:ab_global_and_local
hofee:ab_local_only
..
compare: hofee:2af52c64e26fa77a30b1eae7ecc9457185def23e
Branches Tags
hofee:ab_new_partial_global
hofee:new_partial_global
hofee:ab_global_only
hofee:hemisphere_random
hofee:master
hofee:ab_global_and_partial_global
hofee:ab_global_and_local
hofee:ab_local_only

No commits in common. "60c935749146471c3a5dc0c4fd9523b0ddc63843" and "2af52c64e26fa77a30b1eae7ecc9457185def23e" have entirely different histories.
60c9357491 ... 2af52c64e2

4 changed files with 54 additions and 62 deletions
Show Stats Expand all files Collapse all files

4
configs/local/strategy_generate_config.yaml
View File

@ -28,8 +28,8 @@ runner:
datasets: datasets:
OmniObject3d: OmniObject3d:
#"/media/hofee/data/data/temp_output" #"/media/hofee/data/data/temp_output"
root_dir: "C:\\Document\\Local Project\\nbv_rec\\nbv_reconstruction\\test\\test_sample" root_dir: "/media/hofee/repository/new_full_box_data"
model_dir: "H:\\AI\\Datasets\\scaled_object_meshes" model_dir: "/media/hofee/data/data/scaled_object_meshes"
from: 0 from: 0
to: -1 # -1 means end to: -1 # -1 means end
#output_dir: "/media/hofee/data/data/label_output" #output_dir: "/media/hofee/data/data/label_output"

2
preprocess/preprocessor.py
View File

@ -155,7 +155,7 @@ if __name__ == "__main__":
total = to_idx - from_idx total = to_idx - from_idx
for scene in scene_list[from_idx:to_idx]: for scene in scene_list[from_idx:to_idx]:
start = time.time() start = time.time()
save_scene_data(root, scene, cnt, total, file_type="npy") save_scene_data(root, scene, cnt, total, "npy")
cnt+=1 cnt+=1
end = time.time() end = time.time()
print(f"Time cost: {end-start}") print(f"Time cost: {end-start}")

27
runners/strategy_generator.py
View File

@ -84,38 +84,27 @@ class StrategyGenerator(Runner):
pts_list = [] pts_list = []
scan_points_indices_list = [] scan_points_indices_list = []
non_zero_cnt = 0 non_zero_cnt = 0
for frame_idx in range(frame_num): for frame_idx in range(frame_num):
status_manager.set_progress("generate_strategy", "strategy_generator", "loading frame", frame_idx, frame_num) status_manager.set_progress("generate_strategy", "strategy_generator", "loading frame", frame_idx, frame_num)
pts_path = os.path.join(root,scene_name, "pts", f"{frame_idx}.npy") pts_path = os.path.join(root,scene_name, "target_pts", f"{frame_idx}.txt")
idx_path = os.path.join(root,scene_name, "scan_points_indices", f"{frame_idx}.npy") sampled_point_cloud = np.loadtxt(pts_path)
point_cloud = np.load(pts_path) indices = None # ReconstructionUtil.compute_covered_scan_points(scan_points, display_table_pts)
sampled_point_cloud = PtsUtil.voxel_downsample_point_cloud(point_cloud, voxel_threshold)
indices = np.load(idx_path)
pts_list.append(sampled_point_cloud) pts_list.append(sampled_point_cloud)
scan_points_indices_list.append(indices) scan_points_indices_list.append(indices)
if sampled_point_cloud.shape[0] > 0:
non_zero_cnt += 1
status_manager.set_progress("generate_strategy", "strategy_generator", "loading frame", frame_num, frame_num) status_manager.set_progress("generate_strategy", "strategy_generator", "loading frame", frame_num, frame_num)
seq_num = min(self.seq_num, non_zero_cnt) seq_num = min(self.seq_num, non_zero_cnt)
init_view_list = [] init_view_list = []
idx = 0 for i in range(seq_num):
while len(init_view_list) < seq_num: if pts_list[i].shape[0] < 100:
if pts_list[idx].shape[0] > 100: continue
init_view_list.append(idx) init_view_list.append(i)
idx += 1
seq_idx = 0 seq_idx = 0
import time
for init_view in init_view_list: for init_view in init_view_list:
status_manager.set_progress("generate_strategy", "strategy_generator", "computing sequence", seq_idx, len(init_view_list)) status_manager.set_progress("generate_strategy", "strategy_generator", "computing sequence", seq_idx, len(init_view_list))
start = time.time()
limited_useful_view, _, _ = ReconstructionUtil.compute_next_best_view_sequence_with_overlap(down_sampled_model_pts, pts_list, scan_points_indices_list = scan_points_indices_list,init_view=init_view, limited_useful_view, _, _ = ReconstructionUtil.compute_next_best_view_sequence_with_overlap(down_sampled_model_pts, pts_list, scan_points_indices_list = scan_points_indices_list,init_view=init_view,
threshold=voxel_threshold, soft_overlap_threshold=soft_overlap_threshold, hard_overlap_threshold= hard_overlap_threshold, scan_points_threshold=10, status_info=self.status_info) threshold=voxel_threshold, soft_overlap_threshold=soft_overlap_threshold, hard_overlap_threshold= hard_overlap_threshold, scan_points_threshold=10, status_info=self.status_info)
end = time.time()
print(f"Time: {end-start}")
data_pairs = self.generate_data_pairs(limited_useful_view) data_pairs = self.generate_data_pairs(limited_useful_view)
seq_save_data = { seq_save_data = {
"data_pairs": data_pairs, "data_pairs": data_pairs,

83
utils/reconstruction.py
View File

@ -22,7 +22,29 @@ class ReconstructionUtil:
else: else:
overlap_rate = overlapping_points / new_point_cloud.shape[0] overlap_rate = overlapping_points / new_point_cloud.shape[0]
return overlap_rate return overlap_rate
@staticmethod
def combine_point_with_view_sequence(point_list, view_sequence):
selected_views = []
for view_index, _ in view_sequence:
selected_views.append(point_list[view_index])
return np.vstack(selected_views)
@staticmethod
def compute_next_view_coverage_list(views, combined_point_cloud, target_point_cloud, threshold=0.01):
best_view = None
best_coverage_increase = -1
current_coverage = ReconstructionUtil.compute_coverage_rate(target_point_cloud, combined_point_cloud, threshold)
for view_index, view in enumerate(views):
candidate_views = combined_point_cloud + [view]
down_sampled_combined_point_cloud = PtsUtil.voxel_downsample_point_cloud(candidate_views, threshold)
new_coverage = ReconstructionUtil.compute_coverage_rate(target_point_cloud, down_sampled_combined_point_cloud, threshold)
coverage_increase = new_coverage - current_coverage
if coverage_increase > best_coverage_increase:
best_coverage_increase = coverage_increase
best_view = view_index
return best_view, best_coverage_increase
@staticmethod @staticmethod
def get_new_added_points(old_combined_pts, new_pts, threshold=0.005): def get_new_added_points(old_combined_pts, new_pts, threshold=0.005):
@ -38,70 +60,54 @@ class ReconstructionUtil:
@staticmethod @staticmethod
def compute_next_best_view_sequence_with_overlap(target_point_cloud, point_cloud_list, scan_points_indices_list, threshold=0.01, soft_overlap_threshold=0.5, hard_overlap_threshold=0.7, init_view = 0, scan_points_threshold=5, status_info=None): def compute_next_best_view_sequence_with_overlap(target_point_cloud, point_cloud_list, scan_points_indices_list, threshold=0.01, soft_overlap_threshold=0.5, hard_overlap_threshold=0.7, init_view = 0, scan_points_threshold=5, status_info=None):
selected_views = [init_view] selected_views = [point_cloud_list[init_view]]
combined_point_cloud = point_cloud_list[init_view] combined_point_cloud = np.vstack(selected_views)
history_indices = [scan_points_indices_list[init_view]] history_indices = [scan_points_indices_list[init_view]]
down_sampled_combined_point_cloud = PtsUtil.voxel_downsample_point_cloud(combined_point_cloud,threshold)
max_rec_pts = np.vstack(point_cloud_list) new_coverage = ReconstructionUtil.compute_coverage_rate(target_point_cloud, down_sampled_combined_point_cloud, threshold)
downsampled_max_rec_pts = PtsUtil.voxel_downsample_point_cloud(max_rec_pts, threshold)
max_rec_pts_num = downsampled_max_rec_pts.shape[0]
max_rec_pts_coverage = ReconstructionUtil.compute_coverage_rate(target_point_cloud, downsampled_max_rec_pts, threshold)
new_coverage = ReconstructionUtil.compute_coverage_rate(downsampled_max_rec_pts, combined_point_cloud, threshold)
current_coverage = new_coverage current_coverage = new_coverage
remaining_views = list(range(len(point_cloud_list))) remaining_views = list(range(len(point_cloud_list)))
view_sequence = [(init_view, current_coverage)] view_sequence = [(init_view, current_coverage)]
cnt_processed_view = 0 cnt_processed_view = 0
remaining_views.remove(init_view) remaining_views.remove(init_view)
curr_rec_pts_num = combined_point_cloud.shape[0]
import time
while remaining_views: while remaining_views:
best_view = None best_view = None
best_coverage_increase = -1 best_coverage_increase = -1
best_combined_point_cloud = None
for view_index in remaining_views: for view_index in remaining_views:
if point_cloud_list[view_index].shape[0] == 0: if point_cloud_list[view_index].shape[0] == 0:
continue continue
if selected_views: if selected_views:
new_scan_points_indices = scan_points_indices_list[view_index] new_scan_points_indices = scan_points_indices_list[view_index]
if not ReconstructionUtil.check_scan_points_overlap(history_indices, new_scan_points_indices, scan_points_threshold): if not ReconstructionUtil.check_scan_points_overlap(history_indices, new_scan_points_indices, scan_points_threshold):
overlap_threshold = hard_overlap_threshold overlap_threshold = hard_overlap_threshold
else: else:
overlap_threshold = soft_overlap_threshold overlap_threshold = soft_overlap_threshold
start = time.time()
overlap_rate = ReconstructionUtil.compute_overlap_rate(point_cloud_list[view_index],combined_point_cloud, threshold) combined_old_point_cloud = np.vstack(selected_views)
end = time.time() down_sampled_old_point_cloud = PtsUtil.voxel_downsample_point_cloud(combined_old_point_cloud,threshold)
# print(f"overlap_rate Time: {end-start}") down_sampled_new_view_point_cloud = PtsUtil.voxel_downsample_point_cloud(point_cloud_list[view_index],threshold)
overlap_rate = ReconstructionUtil.compute_overlap_rate(down_sampled_new_view_point_cloud,down_sampled_old_point_cloud, threshold)
if overlap_rate < overlap_threshold: if overlap_rate < overlap_threshold:
continue continue
start = time.time() candidate_views = selected_views + [point_cloud_list[view_index]]
new_combined_point_cloud = np.vstack([combined_point_cloud, point_cloud_list[view_index]]) combined_point_cloud = np.vstack(candidate_views)
new_downsampled_combined_point_cloud = PtsUtil.voxel_downsample_point_cloud(new_combined_point_cloud,threshold) down_sampled_combined_point_cloud = PtsUtil.voxel_downsample_point_cloud(combined_point_cloud,threshold)
new_coverage = ReconstructionUtil.compute_coverage_rate(downsampled_max_rec_pts, new_downsampled_combined_point_cloud, threshold) new_coverage = ReconstructionUtil.compute_coverage_rate(target_point_cloud, down_sampled_combined_point_cloud, threshold)
end = time.time()
#print(f"compute_coverage_rate Time: {end-start}")
coverage_increase = new_coverage - current_coverage coverage_increase = new_coverage - current_coverage
if coverage_increase > best_coverage_increase: if coverage_increase > best_coverage_increase:
best_coverage_increase = coverage_increase best_coverage_increase = coverage_increase
best_view = view_index best_view = view_index
best_combined_point_cloud = new_downsampled_combined_point_cloud
if best_view is not None: if best_view is not None:
if best_coverage_increase <=1e-3: if best_coverage_increase <=3e-3:
break break
selected_views.append(point_cloud_list[best_view])
selected_views.append(best_view)
best_rec_pts_num = best_combined_point_cloud.shape[0]
print(f"Current rec pts num: {curr_rec_pts_num}, Best rec pts num: {best_rec_pts_num}, Max rec pts num: {max_rec_pts_num}")
print(f"Current coverage: {current_coverage}, Best coverage increase: {best_coverage_increase}, Max coverage: {max_rec_pts_coverage}")
curr_rec_pts_num = best_rec_pts_num
combined_point_cloud = best_combined_point_cloud
remaining_views.remove(best_view) remaining_views.remove(best_view)
history_indices.append(scan_points_indices_list[best_view]) history_indices.append(scan_points_indices_list[best_view])
current_coverage += best_coverage_increase current_coverage += best_coverage_increase
@ -117,15 +123,12 @@ class ReconstructionUtil:
else: else:
break break
# ----- Debug Trace ----- #
import ipdb; ipdb.set_trace()
# ------------------------ #
if status_info is not None: if status_info is not None:
sm = status_info["status_manager"] sm = status_info["status_manager"]
app_name = status_info["app_name"] app_name = status_info["app_name"]
runner_name = status_info["runner_name"] runner_name = status_info["runner_name"]
sm.set_progress(app_name, runner_name, "processed view", len(point_cloud_list), len(point_cloud_list)) sm.set_progress(app_name, runner_name, "processed view", len(point_cloud_list), len(point_cloud_list))
return view_sequence, remaining_views, combined_point_cloud return view_sequence, remaining_views, down_sampled_combined_point_cloud
@staticmethod @staticmethod