args = parser.parse_args()
depth_dir = args.depth_dir
for filename in os.listdir(depth_dir):
if not re.match(r'^\d{8}_init.pfm$', filename):
continue
depth_path = os.path.join(depth_dir, filename)
if depth_path.endswith('npy'):
depth_image = np.load(depth_path)
depth_image = np.squeeze(depth_image)
print('value range: ', depth_image.min(), depth_image.max())
plt.imshow(depth_image, 'rainbow')
# plt.show()
elif depth_path.endswith('pfm'):
depth_image = load_pfm(open(depth_path, 'rb'))
ma = np.ma.masked_equal(depth_image, 0.0, copy=False)
print('value range: ', ma.min(), ma.max())
plt.imshow(depth_image, 'rainbow')
# plt.show()
elif depth_path.endswith('dmb'):
depth_image = read_gipuma_dmb(depth_path)
ma = np.ma.masked_equal(depth_image, 0.0, copy=False)
print('value range: ', ma.min(), ma.max())
plt.imshow(depth_image, 'rainbow')
# plt.show()
else:
depth_image = cv2.imread(depth_path)
ma = np.ma.masked_equal(depth_image, 0.0, copy=False)
print('value range: ', ma.min(), ma.max())
plt.imshow(depth_image)
from preprocess import load_pfm
from preprocess import get_normal_from_depth
import os
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('depth_path')
args = parser.parse_args()
path = args.depth_path
depth_files = os.listdir(os.path.join(path, 'depth'))
mask_files = os.listdir(os.path.join(path, 'ground_truth'))
depth_init_files = os.listdir(os.path.join(path, 'init'))
prob_files = os.listdir(os.path.join(path, 'prob'))
for i in range(0, len(depth_files)):
if depth_files[i].endswith('.pfm'):
depth_image = load_pfm(
open(os.path.join(path, 'depth', depth_files[i]), 'rb'))
#normal = get_normal_from_depth(depth_image)
gt_image = load_pfm(
open(os.path.join(path, 'ground_truth', mask_files[i]), 'rb'))
#normal = get_normal_from_depth(gt_image)
gt_image = cv2.resize(gt_image, (0, 0), None, 0.25, 0.25)
init_image = load_pfm(
open(os.path.join(path, 'init', depth_init_files[i]), 'rb'))
prob_image = load_pfm(
open(os.path.join(path, 'prob', prob_files[i]), 'rb'))
mask = gt_image < 0
depth_image[mask] = float('nan')
init_image[mask] = float('nan')
name, ext = os.path.splitext(depth_files[i])
plt.imshow(depth_image, 'rainbow')
enable_prob_filter = True
prob_threshold = 0.8
errors = []
if enable_median_scaling:
ratios = []
for scan in os.listdir(pred_dir):
if scan != 'scan10':
continue
pred_depth_dir = os.path.join(pred_dir, scan, 'depths_mvsnet')
gt_depth_dir = os.path.join(gt_dir, scan + '_train')
for pred_depth_filename in os.listdir(pred_depth_dir):
if not pred_depth_filename.endswith('_init.pfm'):
continue
pred_depth = load_pfm(
open(os.path.join(pred_depth_dir, pred_depth_filename)))
if enable_prob_filter:
pred_prob_filename = pred_depth_filename[:-8] + 'prob.pfm'
pred_prob = load_pfm(
open(os.path.join(pred_depth_dir, pred_prob_filename)))
pred_depth[pred_prob < prob_threshold] = 0
gt_depth_filename = 'depth_map_' + pred_depth_filename[4:8] + '.pfm'
gt_depth = load_pfm(
open(os.path.join(gt_depth_dir, gt_depth_filename)))
mask = np.logical_and(gt_depth > 0, pred_prob > 0.8)
pred_depth = pred_depth[mask]
gt_depth = gt_depth[mask]
if enable_median_scaling:
ratio = np.median(gt_depth) / np.median(pred_depth)
ratios.append(ratio)
import numpy as np
import cv2
import argparse
import matplotlib.pyplot as plt
from preprocess import load_pfm
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('depth_path')
args = parser.parse_args()
depth_path = args.depth_path
if depth_path.endswith('npy'):
depth_image = np.load(depth_path)
depth_image = np.squeeze(depth_image)
print('value range: ', depth_image.min(), depth_image.max())
plt.imshow(depth_image, 'rainbow')
plt.show()
elif depth_path.endswith('pfm'):
depth_image = load_pfm(open(depth_path))
ma = np.ma.masked_equal(depth_image, 0.0, copy=False)
print('value range: ', ma.min(), ma.max())
plt.imshow(depth_image, 'rainbow')
plt.show()
else:
depth_image = cv2.imread(depth_path)
ma = np.ma.masked_equal(depth_image, 0.0, copy=False)
print('value range: ', ma.min(), ma.max())
plt.imshow(depth_image)
plt.show()