def plot_init_potato():
cla = dcp.Classifier(
path_list=['data/potato_fore.png', 'data/potato_back.png'],
kind_list=[0, -1],
core='dtc')
plot = dcp.Plot('data/potato.ply', cla, output_path='output')
return plot
def plot_init_brocoli():
cla = dcp.Classifier(path_list=[
'data/brocoli_train_plant1.ply', 'data/brocoli_train_soil.ply',
'data/brocoli_train_soil_gcp.ply'
],
kind_list=[0, -1, -1],
core='dtc')
plot = dcp.Plot('data/brocoli.ply', cla, output_path='output')
return plot
import easydcp as dcp
cla = dcp.Classifier(
path_list=['training_data/02/fore_rm_r.png', 'training_data/02/back.png'],
kind_list=[0, -1],
core='dtc',
unit='m')
# batch processing
plot_set = ['SP1G2.ply', 'SP4G1.ply'] #ply files must be in working directory
# empty list for batch processing
result_container = []
for plot in plot_set:
plot_class = dcp.Plot(
plot, cla, write_ply=True, unit='m', down_sample=True
) # show_steps=True to display output among calculation to check correct or not
# ---------- auto_segment() --------------
plot_class.pcd_classified = plot_class.remove_noise()
eps, min_points = plot_class.auto_dbscan_args(eps_grids=13, divide=100)
seg = plot_class.dbscan_segment(eps=eps, min_points=min_points)
if len(seg[0]) > 3:
split = plot_class.kmeans_split()
reset_id = plot_class.sort_order(name_by='x', ascending=True)
plot_class.save_segment_result(img_folder='plot_out')
# ----------------------------------------
traits = plot_class.get_traits()
result_container.append(traits)
plot_all = pd.concat(result_container, axis=0).reset_index()
plot_all.to_csv('traits.csv', index=False)
plant_merge_list = ['S04/class[0]-plant2.ply', 'S04/class[0]-plant3.ply']
pcd_list = []
for ply_path in plant_merge_list:
pcd = dcp.read_ply(ply_path)
pcd_list.append(pcd)
m_pcd = dcp.merge_pcd(pcd_list)
#o3d.visualization.draw_geometries([m_pcd])
cla = dcp.Classifier(
path_list=['training_data/fore_rm_y.png', 'training_data/back.png'],
kind_list=[0, -1],
core='dtc')
plot1 = dcp.Plot('S04.ply', cla, show_steps=True)
plant_m = dcp.Plant(plot1, m_pcd, 0, 0)
out_dict = {
'x(m)': [],
'y(m)': [],
'width(m)': [],
'length(m)': [],
'hover_area(m2)': [],
'height(m)': [],
'convex_volume(m3)': [],
'voxel_volume(m3)': []
}
out_dict['x(m)'] = plant_m.center[0]
out_dict['y(m)'] = plant_m.center[1]
core='dtc',
unit='m')
# batch processing
## HZ setting
#plot_set = ['S01.ply', 'S02.ply', 'S03.ply', 'S04.ply', 'S05.ply', 'S06.ply', 'S18.ply', 'S32.ply']
## Alex Setting
plot_set = ['SP1G2.ply', 'SP4G1.ply']
# empty list for batch processing
result_container = []
for plot in plot_set:
# show_steps=True to display output among calculation to check correct or not
plot_class = dcp.Plot(plot,
cla,
write_ply=True,
unit='m',
down_sample=True)
# ---------- auto_segment() --------------
plot_class.pcd_classified = plot_class.remove_noise()
eps, min_points = plot_class.auto_dbscan_args(eps_grids=13, divide=100)
seg = plot_class.dbscan_segment(eps=eps, min_points=min_points)
if len(seg[0]) > 3:
split = plot_class.kmeans_split()
reset_id = plot_class.sort_order(name_by='x', ascending=True)
plot_class.save_segment_result(img_folder='plot_out')
# ----------------------------------------
traits = plot_class.get_traits(container_ht=0.057)
result_container.append(traits)
plot_all = pd.concat(result_container, axis=0).reset_index()
import __init__
import easydcp as dcp
cla = dcp.Classifier(path_list=[
'data/down_sample_train_fore.ply', 'data/down_sample_train_back.ply'
],
kind_list=[0, -1],
core='dtc',
unit='mm')
plot = dcp.Plot('data/down_sample',
cla,
output_path='output',
write_ply=True,
unit='mm')
# start auto_segmentation()
plot.pcd_classified = plot.remove_noise()
eps, min_points = plot.auto_dbscan_args(eps_grids=10)
seg = plot.dbscan_segment(eps=eps, min_points=min_points)
split = plot.kmeans_split()
reset_id = plot.sort_order(name_by='x', ascending=True)
plot.save_segment_result()
# end auto_segmentation()
df = plot.get_traits()
import __init__
import easydcp as dcp
cla = dcp.Classifier(path_list=['data/weed_fore.png', 'data/weed_back.png'],
kind_list=[0, -1],
core='dtc')
plot = dcp.Plot('data/weed.ply', cla, output_path='output')
# start auto_segmentation()
plot.pcd_classified = plot.remove_noise()
eps, min_points = plot.auto_dbscan_args(times=10)
seg = plot.dbscan_segment(eps=eps, min_points=min_points)
split = plot.kmeans_split()
reset_id = plot.sort_order(name_by='x', ascending=True)
plot.save_segment_result()
# end auto_segmentation()
df = plot.get_traits()
import __init__
import easydcp as dcp
cla = dcp.Classifier(
path_list=['training_data/01/fore_rm_y.png', 'training_data/01/back.png'],
kind_list=[0, -1],
core='dtc',
unit='m')
plot1 = dcp.Plot(r'S03.ply', cla, unit='m', write_ply=False, down_sample=True)
# ----------- auto_segmentation() ----------
plot1.pcd_classified = plot1.remove_noise()
# split the shortest axis into 100 parts
eps, min_points = plot1.auto_dbscan_args(eps_grids=13, divide=100)
# the dbscan eps is the length of 10 grids, (10/100=10% of shortest axis in this case)
# the min_points is the mean points of each grids (voxels)
seg = plot1.dbscan_segment(eps=eps, min_points=min_points)
split = plot1.kmeans_split(
) # the algorithm to remove big noise that can not be removed in previous steps (limited by min_points)
reset_id = plot1.sort_order(name_by='x', ascending=True)
plot1.save_segment_result(img_folder='plot_out')
# -------------------------------------------
traits = plot1.get_traits(container_ht=0.057) # size in meter
traits.to_csv('S03/plot1.csv')
import __init__
import easydcp as dcp
import pandas as pd
cla = dcp.Classifier(path_list=['data/brocoli_train_plant1.ply',
'data/brocoli_train_soil.ply',
'data/brocoli_train_soil_gcp.ply'],
kind_list=[0, -1, -1], core='dtc')
plot = dcp.Plot('data/brocoli.ply', cla, output_path='output')
#seg = plot.auto_segmentation()
seg = plot.dbscan_segment(eps=1, min_points=10)
print(seg)
#df = plot.get_traits()
#print(df)