Open yilinliu77 opened 4 years ago
After deleting some codes which are incompatible with our system, we can obtain the GUI of your program now but fail to generate the next view from the original path. After checking in the source code, we found that the ray cast always returns blank when it attempts to generate new candidate view. But the voxel seems fine in the GUI.
Can you give us some advice about it?
We found there are something not compatible with the binary ouput of colmap 3.6, we solved it. Fortunately, we can generate candidate views now. However, the generated views are concentrated in part area of the scene, even though it cost several hours to generate them. Could you please give us some suggestions to solve this problem? Are there any parameters that maybe cause this?
Here we show the screen shot of our result, and also attach the Configuration file “viewpoint_planner.cfg” which has our parameters.
We would be very grateful if you can help. Thanks very much.
Here is our config file
[viewpoint_planner]
rng_seed = 0
virtual_camera_scale = 0.05
#virtual_camera_scale = 0.15
viewpoint_exploration_step = 10
raycast_max_range = 120
# drone_extent_x = 3
# drone_extent_y = 3
# drone_extent_z = 3
sampling_roi_factor = 2
pose_sample_num_trials = 100
pose_sample_min_radius = 2
pose_sample_max_radius = 5
# viewpoint_sample_count = 10
viewpoint_min_voxel_count = 50
viewpoint_min_information = 10
viewpoint_discard_dist_knn = 20
viewpoint_discard_dist_thres_square = 4
viewpoint_discard_dist_count_thres = 3
viewpoint_motion_max_neighbors = 20
viewpoint_motion_max_dist_square = 100
# viewpoint_motion_min_connections = 3
viewpoint_motion_densification_max_depth = 8
viewpoint_path_branches = 1
#viewpoint_path_branches = 20
viewpoint_path_initial_distance = 6
objective_parameter_alpha = 0
objective_parameter_beta = 0
viewpoint_path_2opt_max_k_length = 25
#viewpoint_graph_filename = viewpoint_graph.bs
#viewpoint_graph_filename = viewpoint_graph_with_motions.bs
num_sampled_poses = 20
num_planned_viewpoints = 1
#num_sampled_poses = 250
#num_planned_viewpoints = 25
#num_sampled_poses = 1000
#num_planned_viewpoints = 50
[motion_planner]
max_motion_range = 5
max_time_per_solve = 0.01
max_iterations_per_solve = 1000
[viewpoint_planner.data]
dense_reconstruction_path = /home/szu/NY-1-colmap-workspace
poisson_mesh_filename = /home/szu/NY-1-colmap-workspace/meshed-delaunay_txt.ply
raw_octree_filename = /home/szu/quad_origin/build/ny_1.ot
dense_points_filename = /home/szu/NY-1-colmap-workspace/fused_txt.ply
#raw_octree_filename = /home/bhepp/Desktop/restroom/dense_far3.ot
#octree_filename =
#bvh_filename =
#distance_field_filename =
#regenerate_augmented_octree = True
#regenerate_bvh_tree = True
#regenerate_distance_field = True
grid_dimension = 200
distance_field_cutoff = 5
bvh_bbox_min_x = -40
bvh_bbox_min_y = -40
bvh_bbox_min_z = -10
bvh_bbox_max_x = +40
bvh_bbox_max_y = +50
bvh_bbox_max_z = +100
roi_bbox_min_x = -24
roi_bbox_min_y = -30
roi_bbox_min_z = 0
roi_bbox_max_x = +24
roi_bbox_max_y = +44
roi_bbox_max_z = 25
roi_falloff_distance = 10.0
@whatseven Are you able to make this code run your own dataset ?
@whatseven same question as @szx0112 - could you get the system working or did you abandoned the project?
Could anyone provide some informations to help me build this program. I have tried so many combinations but failed to build it.
I have tried:
* ubuntu14 + cuda7.5 + gcc4.8 (cmake failed with qt5) * ubuntu16 + cuda7.5 + gcc4.8 (make error with some functions in octomap) * ubuntu18 + cuda10.2 + gcc8 (make error with some cuda functions)
Especially, I want to know:
* ubuntu 14/16/18? * version of gcc * version of cuda * version of zed (if needed?)
Also, I list some of the dependency of this program, hope it would be help for anyone else.
sudo apt install -y libglew-dev libglfw-dev libboost-dev libboost-filesystem-dev libboost-serialization-dev libboost-program-options-dev libopencv-dev libtbb-dev cmake git libtclap-dev libompl-dev libqglviewer-dev liboctomap-dev liboctovis-dev libqt5websockets5-dev libgstreamer1.0-dev libglm-dev
I have the same problem and cannot execute it successfully during make. How did you solve it?
Could anyone provide some informations to help me build this program. I have tried so many combinations but failed to build it.
I have tried:
Especially, I want to know:
Also, I list some of the dependency of this program, hope it would be help for anyone else.
sudo apt install -y libglew-dev libglfw-dev libboost-dev libboost-filesystem-dev libboost-serialization-dev libboost-program-options-dev libopencv-dev libtbb-dev cmake git libtclap-dev libompl-dev libqglviewer-dev liboctomap-dev liboctovis-dev libqt5websockets5-dev libgstreamer1.0-dev libglm-dev