basalt/scripts/eval_full/gen_results.py

108 lines
2.6 KiB
Python
Executable File

#!/usr/bin/env python3
#
# BSD 3-Clause License
#
# This file is part of the Basalt project.
# https://gitlab.com/VladyslavUsenko/basalt.git
#
# Copyright (c) 2019-2021, Vladyslav Usenko and Nikolaus Demmel.
# All rights reserved.
#
import os
import sys
import json
datasets = ['Seq.', 'MH_01_easy', 'MH_02_easy', 'MH_03_medium', 'MH_04_difficult',
'MH_05_difficult', 'V1_01_easy', 'V1_02_medium',
'V1_03_difficult', 'V2_01_easy', 'V2_02_medium']
# Other results.
vio = {
'ate' : ['VIO RMS ATE [m]'],
'time' : ['VIO Time [s]'],
'num_frames' : ['VIO Num. Frames']
}
mapping = {
'ate' : ['MAP RMS ATE [m]'],
'time' : ['MAP Time [s]'],
'num_frames' : ['MAP Num. KFs']
}
pose_graph = {
'ate' : ['PG RMS ATE [m]'],
'time' : ['PG Time [s]'],
'num_frames' : ['PG Num. KFs']
}
pure_ba = {
'ate' : ['PG RMS ATE [m]'],
'time' : ['PG Time [s]'],
'num_frames' : ['PG Num. KFs']
}
out_dir = sys.argv[1]
def load_data(x, prefix, key):
fname = out_dir + '/' + prefix + '_' + key
if os.path.isfile(fname):
with open(fname, 'r') as f:
j = json.load(f)
res = round(j['rms_ate'], 3)
x['ate'].append(float(res))
x['time'].append(round(j['exec_time_ns']*1e-9, 3))
x['num_frames'].append(j['num_frames'])
else:
x['ate'].append(float('Inf'))
x['time'].append(float('Inf'))
x['num_frames'].append(float('Inf'))
for key in datasets[1:]:
load_data(vio, 'vio', key)
load_data(mapping, 'mapper', key)
load_data(pose_graph, 'mapper_no_weights', key)
load_data(pure_ba, 'mapper_no_factors', key)
row_format ="{:>17}" + "{:>13}" * (len(datasets)-1)
datasets_short = [x[:5] for x in datasets]
print('\nVisual-Inertial Odometry')
print(row_format.format(*datasets_short))
print(row_format.format(*vio['ate']))
#print(row_format.format(*vio['time']))
print(row_format.format(*vio['num_frames']))
print('\nVisual-Inertial Mapping')
print(row_format.format(*datasets_short))
print(row_format.format(*mapping['ate']))
#print(row_format.format(*mapping['time']))
print(row_format.format(*mapping['num_frames']))
print('\nPose-Graph optimization (Identity weights for all factors)')
print(row_format.format(*datasets_short))
print(row_format.format(*pose_graph['ate']))
#print(row_format.format(*pose_graph['time']))
print(row_format.format(*pose_graph['num_frames']))
print('\nPure BA optimization (no factors from the recovery used)')
print(row_format.format(*datasets_short))
print(row_format.format(*pure_ba['ate']))
#print(row_format.format(*pure_ba['time']))
print(row_format.format(*pure_ba['num_frames']))