Skip to content

L
lwir-nn
Commit 4b02b283 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

more versions, adding more images in the results plot

parent 466ed6b1
Hide whitespace changes
Inline Side-by-side
Showing with 668 additions and 238 deletions
explore_data15.py
View file @ 4b02b283
...@@ -1755,7 +1755,7 @@ if __name__ == "__main__": ...@@ -1755,7 +1755,7 @@ if __name__ == "__main__":
test_corrs = [] test_corrs = []
#1527257933_150165-ML_DATA-32B-O-FZ0.05-MAIN-RND2.00000.tiff #1527257933_150165-ML_DATA-32B-O-FZ0.05-MAIN-RND2.00000.tiff
#/home/eyesis/x3d_data/data_sets/test_mlr32_18a/1527257933_150165/v04/mlr32_18c/1527257933_150165-ML_DATA-32B-O-FZ0.05-MAIN.tiff #/home/eyesis/x3d_data/data_sets/test_mlr32_18a/1527257933_150165/v04/mlr32_18c/1527257933_150165-ML_DATA-32B-O-FZ0.05-MAIN.tiff
'''
test_sets = [ test_sets = [
"/data_ssd/lwir_sets/lwir_test2/1562390202_933097/v01/ml32", # andrey /empty "/data_ssd/lwir_sets/lwir_test2/1562390202_933097/v01/ml32", # andrey /empty
"/data_ssd/lwir_sets/lwir_test2/1562390225_269784/v01/ml32", # andrey /empty "/data_ssd/lwir_sets/lwir_test2/1562390225_269784/v01/ml32", # andrey /empty
...@@ -1777,8 +1777,19 @@ if __name__ == "__main__": ...@@ -1777,8 +1777,19 @@ if __name__ == "__main__":
"/data_ssd/lwir_sets/lwir_test3/1562390409_661607/v01/ml32", # lena, 2 far moving cars "/data_ssd/lwir_sets/lwir_test3/1562390409_661607/v01/ml32", # lena, 2 far moving cars
"/data_ssd/lwir_sets/lwir_test3/1562390435_873048/v01/ml32", # 2 parked cars, lena "/data_ssd/lwir_sets/lwir_test3/1562390435_873048/v01/ml32", # 2 parked cars, lena
"/data_ssd/lwir_sets/lwir_test3/1562390456_842237/v01/ml32", # near trees "/data_ssd/lwir_sets/lwir_test3/1562390456_842237/v01/ml32", # near trees
"/data_ssd/lwir_sets/lwir_test3/1562390460_261151/v01/ml32"] # near trees, olga "/data_ssd/lwir_sets/lwir_test3/1562390460_261151/v01/ml32", # near trees, olga
]
'''
test_sets = [
"/data_ssd/lwir_sets/lwir_test6/1562390317_693673/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test6/1562390318_833313/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test6/1562390326_354823/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test6/1562390331_483132/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test6/1562390333_192523/v01/ml32", # lena
]
#Parameters to generate neighbors data. Set radius to 0 to generate single-tile #Parameters to generate neighbors data. Set radius to 0 to generate single-tile
TEST_SAME_LENGTH_AS_TRAIN = False # True # make test to have same number of entries as train ones TEST_SAME_LENGTH_AS_TRAIN = False # True # make test to have same number of entries as train ones
FIXED_TEST_LENGTH = 102 # None # put number of test scenes to output (used when making test only from few or single test file FIXED_TEST_LENGTH = 102 # None # put number of test scenes to output (used when making test only from few or single test file
......
explore_data16.py
View file @ 4b02b283
...@@ -1657,27 +1657,28 @@ if __name__ == "__main__": ...@@ -1657,27 +1657,28 @@ if __name__ == "__main__":
test_corrs = [] test_corrs = []
test_sets = [ test_sets = [
"/data_ssd/lwir_sets/lwir_test2/1562390202_933097/v01/ml32", # andrey /empty "/data_ssd/lwir_sets/lwir_test2/1562390202_933097/v01/ml32b", # andrey /empty
"/data_ssd/lwir_sets/lwir_test2/1562390225_269784/v01/ml32", # andrey /empty "/data_ssd/lwir_sets/lwir_test2/1562390225_269784/v01/ml32b", # andrey /empty
"/data_ssd/lwir_sets/lwir_test2/1562390225_839538/v01/ml32", # andrey /empty "/data_ssd/lwir_sets/lwir_test2/1562390225_839538/v01/ml32b", # andrey /empty
"/data_ssd/lwir_sets/lwir_test2/1562390243_047919/v01/ml32", # 2 trees "/data_ssd/lwir_sets/lwir_test2/1562390243_047919/v01/ml32b", # 2 trees
"/data_ssd/lwir_sets/lwir_test2/1562390251_025390/v01/ml32", # empty space
"/data_ssd/lwir_sets/lwir_test2/1562390257_977146/v01/ml32", # first 3
"/data_ssd/lwir_sets/lwir_test2/1562390260_370347/v01/ml32", # all 3
"/data_ssd/lwir_sets/lwir_test2/1562390260_940102/v01/ml32", # all 3
"/data_ssd/lwir_sets/lwir_test6/1562390317_693673/v01/ml32", # andrey + olga "/data_ssd/lwir_sets/lwir_test6/1562390251_025390/v01/ml32b", # empty space
"/data_ssd/lwir_sets/lwir_test6/1562390318_833313/v01/ml32", # andrey + olga "/data_ssd/lwir_sets/lwir_test6/1562390257_977146/v01/ml32b", # first 3
"/data_ssd/lwir_sets/lwir_test6/1562390326_354823/v01/ml32", # andrey + olga "/data_ssd/lwir_sets/lwir_test6/1562390260_370347/v01/ml32b", # all 3
"/data_ssd/lwir_sets/lwir_test6/1562390331_483132/v01/ml32", # andrey + olga "/data_ssd/lwir_sets/lwir_test2/1562390260_940102/v01/ml32b", # all 3
"/data_ssd/lwir_sets/lwir_test6/1562390333_192523/v01/ml32", # lena
"/data_ssd/lwir_sets/lwir_test3/1562390402_254007/v01/ml32", # near moving car "/data_ssd/lwir_sets/lwir_test6/1562390317_693673/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test3/1562390407_382326/v01/ml32", # near moving car "/data_ssd/lwir_sets/lwir_test6/1562390318_833313/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test3/1562390409_661607/v01/ml32", # lena, 2 far moving cars "/data_ssd/lwir_sets/lwir_test6/1562390326_354823/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test3/1562390435_873048/v01/ml32", # 2 parked cars, lena "/data_ssd/lwir_sets/lwir_test6/1562390331_483132/v01/ml32", # andrey + olga
"/data_ssd/lwir_sets/lwir_test3/1562390456_842237/v01/ml32", # near trees "/data_ssd/lwir_sets/lwir_test6/1562390333_192523/v01/ml32", # lena
"/data_ssd/lwir_sets/lwir_test3/1562390460_261151/v01/ml32"] # near trees, olga
"/data_ssd/lwir_sets/lwir_test6/1562390402_254007/v01/ml32b", # near moving car
"/data_ssd/lwir_sets/lwir_test6/1562390407_382326/v01/ml32b", # near moving car
"/data_ssd/lwir_sets/lwir_test6/1562390409_661607/v01/ml32b", # lena, 2 far moving cars
"/data_ssd/lwir_sets/lwir_test6/1562390435_873048/v01/ml32b", # 2 parked cars, lena
"/data_ssd/lwir_sets/lwir_test6/1562390456_842237/v01/ml32b", # near trees
"/data_ssd/lwir_sets/lwir_test6/1562390460_261151/v01/ml32b"] # near trees, olga
#Parameters to generate neighbors data. Set radius to 0 to generate single-tile #Parameters to generate neighbors data. Set radius to 0 to generate single-tile
TEST_SAME_LENGTH_AS_TRAIN = False # True # make test to have same number of entries as train ones TEST_SAME_LENGTH_AS_TRAIN = False # True # make test to have same number of entries as train ones
......
nn_ds_neibs31.py
View file @ 4b02b283
...@@ -3,7 +3,7 @@ __copyright__ = "Copyright 2018-2019, Elphel, Inc." ...@@ -3,7 +3,7 @@ __copyright__ = "Copyright 2018-2019, Elphel, Inc."
__license__ = "GPL-3.0+" __license__ = "GPL-3.0+"
__email__ = "andrey@elphel.com" __email__ = "andrey@elphel.com"
#python3 nn_ds_neibs31.py /data_ssd/lwir_sets/conf/qcstereo_lwir05.xml /data_ssd/lwir_sets/ #python3 nn_ds_neibs31.py /data_ssd/lwir_sets/conf/qcstereo_lwir21.xml /data_ssd/lwir_sets/
#tensorboard --logdir="nn_ds_neibs30_graph13-9RNSWLAM0.5SLAM0.1SCLP0.2_nG_nI_HF_CP0.3_S0.03" --port=7001 #tensorboard --logdir="nn_ds_neibs30_graph13-9RNSWLAM0.5SLAM0.1SCLP0.2_nG_nI_HF_CP0.3_S0.03" --port=7001
import os import os
......
nn_eval_lwir.py
View file @ 4b02b283
...@@ -16,7 +16,7 @@ import sys ...@@ -16,7 +16,7 @@ import sys
import imagej_tiffwriter import imagej_tiffwriter
import time import time
import imagej_tiff as ijt
import matplotlib.pyplot as plt import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages from matplotlib.backends.backend_pdf import PdfPages
import qcstereo_functions as qsf import qcstereo_functions as qsf
...@@ -27,7 +27,8 @@ import numpy as np ...@@ -27,7 +27,8 @@ import numpy as np
qsf.TIME_START = time.time() qsf.TIME_START = time.time()
qsf.TIME_LAST = qsf.TIME_START qsf.TIME_LAST = qsf.TIME_START
IMG_WIDTH = 20 # 324 # tiles per image row Defined in config #IMG_WIDTH = 20 # 324 # tiles per image row Defined in config
#IMG_HEIGHT = 15 # 324 # tiles per image row Defined in config
DEBUG_LEVEL= 1 DEBUG_LEVEL= 1
...@@ -40,6 +41,12 @@ try: ...@@ -40,6 +41,12 @@ try:
root_dir = sys.argv[2] root_dir = sys.argv[2]
except IndexError: except IndexError:
root_dir = os.path.dirname(conf_file) root_dir = os.path.dirname(conf_file)
try:
modes = [sys.argv[3]] # train, infer
except IndexError:
modes = ['train']
print ("Configuration file: " + conf_file) print ("Configuration file: " + conf_file)
parameters, dirs, files, dbg_parameters = qsf.parseXmlConfig(conf_file, root_dir) parameters, dirs, files, dbg_parameters = qsf.parseXmlConfig(conf_file, root_dir)
...@@ -53,6 +60,8 @@ if not "SLOSS_CLIP" in parameters: ...@@ -53,6 +60,8 @@ if not "SLOSS_CLIP" in parameters:
""" """
Defined in config file Defined in config file
""" """
IMG_WIDTH = None # 20 # 324 # tiles per image row Defined in config
IMG_HEIGHT = None # 15 # 324 # tiles per image row Defined in config
TILE_SIDE, TILE_LAYERS, TWO_TRAINS, NET_ARCH1, NET_ARCH2 = [None]*5 TILE_SIDE, TILE_LAYERS, TWO_TRAINS, NET_ARCH1, NET_ARCH2 = [None]*5
ABSOLUTE_DISPARITY,SYM8_SUB, WLOSS_LAMBDA, SLOSS_LAMBDA, SLOSS_CLIP = [None]*5 ABSOLUTE_DISPARITY,SYM8_SUB, WLOSS_LAMBDA, SLOSS_LAMBDA, SLOSS_CLIP = [None]*5
SPREAD_CONVERGENCE, INTER_CONVERGENCE, HOR_FLIP, DISP_DIFF_CAP, DISP_DIFF_SLOPE = [None]*5 SPREAD_CONVERGENCE, INTER_CONVERGENCE, HOR_FLIP, DISP_DIFF_CAP, DISP_DIFF_SLOPE = [None]*5
...@@ -111,14 +120,19 @@ qsf.prepareFiles(dirs, files, suffix = SUFFIX) ...@@ -111,14 +120,19 @@ qsf.prepareFiles(dirs, files, suffix = SUFFIX)
CONF_MAX = 0.7 CONF_MAX = 0.7
ERR_AMPL = 0.3 ERR_AMPL = 0.4 # 0.3
TIGHT_TOP = 0.95 TIGHT_TOP = 0.95
TIGHT_HPAD = 1.0 TIGHT_HPAD = 1.0
TIGHT_WPAD = 1.0 TIGHT_WPAD = 1.0
FIGSIZE = [8.5,11.0] FIGSIZE = [8.5,11.0]
WOI_COLOR = "red" WOI_COLOR = "red"
X_COLOR = "grey"
X_NEIBS = False
TRANSPARENT = True # for export TRANSPARENT = True # for export
#dbg_parameters #dbg_parameters
def get_fig_params(disparity_ranges): def get_fig_params(disparity_ranges):
fig_params = [] fig_params = []
...@@ -138,7 +152,14 @@ def get_fig_params(disparity_ranges): ...@@ -138,7 +152,14 @@ def get_fig_params(disparity_ranges):
return fig_params return fig_params
#try: #try:
fig_params = get_fig_params(dbg_parameters['disparity_ranges']) #fig_params = get_fig_params(dbg_parameters['disparity_ranges'])
extra_path = os.path.join(root_dir, dbg_parameters['extra'])
eo_width = dbg_parameters['eo_params']['width']
eo_height = dbg_parameters['eo_params']['height']
eo_woi = dbg_parameters['eo_params']['woi'] # (x,y,width, height)
eo_disparity_scale = 1.0/dbg_parameters['eo_params']['disparity_scale'] # 14.2
image_sets = dbg_parameters['extra_paths'] # list of dictionaries
pass pass
...@@ -162,10 +183,15 @@ index_gt = 2 ...@@ -162,10 +183,15 @@ index_gt = 2
index_gt_weight = 3 index_gt_weight = 3
index_heur_err = 7 index_heur_err = 7
index_nn_err = 6 index_nn_err = 6
index_mm = 8 # max-min index_fgbg_sngl = 10
index_log = 9 index_fgbg_neib = 11
index_bad = 10
index_num_neibs = 11 index_mm = 23 # 8 # max-min
index_log = 24 # 9
index_bad = 25 # 10
index_num_neibs = 26 # 11
index_fgbg = [index_fgbg_sngl,index_fgbg_neib][X_NEIBS]
""" """
Debugging high 9-tile variations, removing error for all tiles with lower difference between max and min Debugging high 9-tile variations, removing error for all tiles with lower difference between max and min
""" """
...@@ -181,7 +207,21 @@ if not 'show' in FIGS_SAVESHOW: ...@@ -181,7 +207,21 @@ if not 'show' in FIGS_SAVESHOW:
#for mode in ['train','infer']: #for mode in ['train','infer']:
#for mode in ['infer']: #for mode in ['infer']:
for mode in ['train']: def cross_out(
plt,
cross_out_mask):
height = cross_out_mask.shape[0]
width = cross_out_mask.shape[1]
for row in range (height):
for col in range(width):
if cross_out_mask[row,col]:
xdata = [col-0.3, col+0.3]
ydata = [row-0.3, row+0.3]
plt.plot(xdata,ydata,color=X_COLOR)
ydata = [row+0.3, row-0.3]
plt.plot(xdata,ydata,color=X_COLOR)
for mode in modes: # ['train']:
figs = [] figs = []
ffiles = [] # no ext ffiles = [] # no ext
def setlimsxy(lim_xy): def setlimsxy(lim_xy):
...@@ -189,218 +229,106 @@ for mode in ['train']: ...@@ -189,218 +229,106 @@ for mode in ['train']:
plt.xlim(min(lim_xy[:2]),max(lim_xy[:2])) plt.xlim(min(lim_xy[:2]),max(lim_xy[:2]))
plt.ylim(max(lim_xy[2:]),min(lim_xy[2:])) plt.ylim(max(lim_xy[2:]),min(lim_xy[2:]))
cumul_weights = None cumul_weights = None
cmap_disp=plt.get_cmap('viridis') # ('cividis')
for nfile, fpars in enumerate(fig_params): cmap_diff=plt.get_cmap('coolwarm') #('seismic') # ('viridis')
if not fpars is None: for nfile, img_pars in enumerate(image_sets):
if not img_pars is None:
img_file = files['result'][nfile] img_file = files['result'][nfile]
if mode == 'infer': if mode == 'infer':
img_file = img_file.replace('.npy','-infer.npy') img_file = img_file.replace('.npy','-infer.npy')
""" print ("Processing image set: "+img_file)
try:
# data,_ = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=True)
# data,_ = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
data,labels = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
except:
print ("Image file does not exist:", img_file)
continue
"""
pass
data,labels = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3) data,labels = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
if True: #TIFF_ONLY: cross_out_mask = data[...,index_fgbg] < 0.5 #data.shape = (15,20,27)
# for subindex, rng in enumerate(fpars['ranges']):
tiff_path = img_file.replace('.npy','-test.tiff')
data = data.transpose(2,0,1)
print("Saving results to TIFF: "+tiff_path)
imagej_tiffwriter.save(tiff_path,data,labels=labels)
"""
Calculate histograms
"""
err_heur2 = data[index_heur_err]*data[index_heur_err]
err_nn2 = data[index_nn_err]* data[index_nn_err]
diff_log2 = data[index_log]* data[index_log]
weights = (
(data[index_gt] < max_disp) &
(err_heur2 < max_target_err2) &
(data[index_bad] < max_bad) &
(data[index_gt_weight] >= min_strength) &
(data[index_num_neibs] >= min_neibs)&
#max_log_to_mm = 0.5 # difference between center average and center should be under this fraction of max-min (0 - disables feature)
(data[index_log] < max_log_to_mm * np.sqrt(data[index_mm]) )
).astype(data.dtype) # 0.0/1.1
#max_disp
#max_target_err
if use_gt_weights:
weights *= data[index_gt_weight]
mm = data[index_mm]
weh = np.nan_to_num(weights*err_heur2)
wen = np.nan_to_num(weights*err_nn2)
wel = np.nan_to_num(weights*diff_log2)
hist_weights,bin_vals = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = weights, density = False)
hist_err_heur2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = weh, density = False)
hist_err_nn2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = wen, density = False)
hist_diff_log2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = wel, density = False)
if cumul_weights is None:
cumul_weights = hist_weights
cumul_err_heur2 = hist_err_heur2
cumul_err_nn2 = hist_err_nn2
cumul_diff_log2 = hist_diff_log2
else:
cumul_weights += hist_weights
cumul_err_heur2 += hist_err_heur2
cumul_err_nn2 += hist_err_nn2
cumul_diff_log2 += hist_diff_log2
hist_err_heur2 = np.nan_to_num(hist_err_heur2/hist_weights)
hist_err_nn2 = np.nan_to_num(hist_err_nn2/hist_weights)
hist_gain2 = np.nan_to_num(hist_err_heur2/hist_err_nn2)
hist_gain = np.sqrt(hist_gain2)
hist_diff_log2 = np.nan_to_num(hist_diff_log2/hist_weights)
print("hist_err_heur2", end = " ")
print(np.sqrt(hist_err_heur2))
print("hist_err_nn2", end = " ")
print(np.sqrt(hist_err_nn2))
print("hist_gain", end = " ")
print(hist_gain)
print("hist_diff_log2", end = " ")
print(np.sqrt(hist_diff_log2))
if min_diff> 0.0:
pass
good = (mm > min_diff).astype(mm.dtype)
good /= good # good - 1, bad - nan
data[index_heur_err] *= good
data[index_nn_err] *= good
data = data.transpose(1,2,0)
if TIFF_ONLY: lim_val = img_pars['range'] # rng['lim_val']
continue lim_val[0] -= ERR_AMPL
lim_xy = [-0.5, IMG_WIDTH - 0.5, -0.5, IMG_HEIGHT - 0.5] # rng['lim_xy']
for subindex, rng in enumerate(fpars['ranges']):
lim_val = rng['lim_val']
lim_xy = rng['lim_xy']
fig = plt.figure(figsize=FIGSIZE)
fig.canvas.set_window_title(fpars['name'])
fig.suptitle(fpars['name'])
ax_conf=plt.subplot(322)
ax_conf.set_title("Ground truth confidence")
# fig.suptitle("Groud truth confidence")
plt.imshow(data[...,qsf.GT_CONF], vmin=0, vmax=CONF_MAX, cmap='gray')
if not lim_xy is None:
pass # show frame
xdata=[min(lim_xy[:2]),max(lim_xy[:2]),max(lim_xy[:2]),min(lim_xy[:2]),min(lim_xy[:2])]
ydata=[min(lim_xy[2:]),min(lim_xy[2:]),max(lim_xy[2:]),max(lim_xy[2:]),min(lim_xy[2:])]
plt.plot(xdata,ydata,color=WOI_COLOR)
# setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
ax_gtd=plt.subplot(321)
ax_gtd.set_title("Ground truth disparity map")
plt.imshow(data[...,qsf.GT_DISP], vmin=lim_val[0], vmax=lim_val[1])
setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
ax_hed=plt.subplot(323)
ax_hed.set_title("Heuristic disparity map")
plt.imshow(data[...,qsf.HEUR_NAN], vmin=lim_val[0], vmax=lim_val[1])
setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
ax_nnd=plt.subplot(325) #start new image page
ax_nnd.set_title("Network disparity output") fig = plt.figure(figsize=FIGSIZE)
plt.imshow(data[...,qsf.NN_NAN], vmin=lim_val[0], vmax=lim_val[1]) fig.canvas.set_window_title(img_pars['title'])
setlimsxy(lim_xy) fig.suptitle(img_pars['title'])
plt.colorbar(orientation='vertical') # location='bottom')
ax_hee=plt.subplot(324)
ax_hee.set_title("Heuristic disparity error")
plt.imshow(data[...,qsf.HEUR_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL)
setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
ax_nne=plt.subplot(326)
ax_nne.set_title("Network disparity error")
plt.imshow(data[...,qsf.NN_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL)
setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
plt.tight_layout(rect =[0,0,1,TIGHT_TOP], h_pad = TIGHT_HPAD, w_pad = TIGHT_WPAD) # Create EO DSI image
# load tiff image
figs.append(fig)
fb_noext = os.path.splitext(os.path.basename(img_file))[0]# img_ds_main = ijt.imagej_tiff(os.path.join(extra_path,img_pars['dsi_path'] ))
if subindex > 0: ds_main = img_ds_main.image[...,img_pars['dsi_slice']] * eo_disparity_scale
if subindex < 10: ds_main = np.maximum(ds_main, lim_val[0])
fb_noext+="abcdefghi"[subindex-1] ds_main = np.minimum(ds_main, lim_val[1])
else: ax_conf=plt.subplot(322)
fb_noext+="-"+str(subindex) ax_conf.set_title("Hi-res camera disparity map")
ffiles.append(fb_noext) plt.imshow(ds_main, vmin=lim_val[0], vmax=lim_val[1], cmap=cmap_disp)
pass setlimsxy([-0.5, eo_width-0.5, -0.5, eo_height - 0.5])
if True: if not eo_woi is None:
cumul_err_heur2 = np.nan_to_num(cumul_err_heur2/cumul_weights) pass # show frame
cumul_err_nn2 = np.nan_to_num(cumul_err_nn2/cumul_weights) xdata=[eo_woi['x'], eo_woi['x'] + eo_woi['width'], eo_woi['x'] + eo_woi['width'], eo_woi['x'], eo_woi['x']]
cumul_gain2 = np.nan_to_num(cumul_err_heur2/cumul_err_nn2) ydata=[eo_woi['y'], eo_woi['y'], eo_woi['y'] + eo_woi['height'], eo_woi['y'] + eo_woi['height'], eo_woi['y']]
cumul_gain = np.sqrt(cumul_gain2) plt.plot(xdata,ydata,color=WOI_COLOR)
cumul_diff_log2 = np.nan_to_num(cumul_diff_log2/cumul_weights)
print("cumul_weights", end = " ") plt.colorbar(orientation='vertical') # location='bottom')
print(cumul_weights) '''
print("cumul_err_heur", end = " ") # Ground truth confidence - to be replaced
print(np.sqrt(cumul_err_heur2)) ax_conf=plt.subplot(322)
print("cumul_err_nn", end = " ") ax_conf.set_title("Ground truth confidence")
print(np.sqrt(cumul_err_nn2)) plt.imshow(data[...,qsf.GT_CONF], vmin=0, vmax=CONF_MAX, cmap='gray')
print("cumul_gain", end = " ") if not lim_xy is None:
print(cumul_gain) pass # show frame
print("cumul_diff_log2", end = " ") xdata=[min(lim_xy[:2]),max(lim_xy[:2]),max(lim_xy[:2]),min(lim_xy[:2]),min(lim_xy[:2])]
print(np.sqrt(cumul_diff_log2)) ydata=[min(lim_xy[2:]),min(lim_xy[2:]),max(lim_xy[2:]),max(lim_xy[2:]),min(lim_xy[2:])]
fig, ax1 = plt.subplots() plt.plot(xdata,ydata,color=WOI_COLOR)
ax1.set_xlabel('3x3 tiles ground truth disparity max-min (pix)') plt.colorbar(orientation='vertical') # location='bottom')
ax1.set_ylabel('RMSE\n(pix)', color='black', rotation='horizontal') '''
ax1.yaxis.set_label_coords(-0.045,0.92)
ax_gtd=plt.subplot(321)
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_err_nn2), 'tab:red',label="network disparity RMSE") ax_gtd.set_title("Ground truth disparity map")
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_err_heur2), 'tab:green',label="heuristic disparity RMSE") plt.imshow(data[...,qsf.GT_DISP], vmin=lim_val[0], vmax=lim_val[1], cmap=cmap_disp)
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_diff_log2), 'tab:cyan',label="ground truth LoG") setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
ax1.tick_params(axis='y', labelcolor='black') plt.colorbar(orientation='vertical') # location='bottom')
ax2 = ax1.twinx() # instantiate a second axes that shares the same x-axis ax_hed=plt.subplot(323)
ax2.set_ylabel('weight', color='black', rotation='horizontal') # we already handled the x-label with ax1 ax_hed.set_title("Heuristic disparity map")
ax2.yaxis.set_label_coords(1.06,1.0) plt.imshow(data[...,qsf.HEUR_NAN], vmin=lim_val[0], vmax=lim_val[1], cmap=cmap_disp)
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
ax2.plot(bin_vals[0:-1], cumul_weights,color='grey',dashes=[6, 2],label='weights = n_tiles * gt_confidence')
ax1.legend(loc="upper left", bbox_to_anchor=(0.2,1.0))
ax2.legend(loc="lower right", bbox_to_anchor=(1.0,0.1))
""" ax_nnd=plt.subplot(325)
ax_nnd.set_title("Network disparity output")
fig = plt.figure(figsize=FIGSIZE) plt.imshow(data[...,qsf.NN_NAN], vmin=lim_val[0], vmax=lim_val[1], cmap=cmap_disp)
fig.canvas.set_window_title('Cumulative') setlimsxy(lim_xy)
fig.suptitle('Difference to GT') cross_out(plt, cross_out_mask)
# ax_conf=plt.subplot(322) plt.colorbar(orientation='vertical') # location='bottom')
ax_conf=plt.subplot(211)
ax_conf.set_title("RMS vs max9-min9") ax_hee=plt.subplot(324)
plt.plot(bin_vals[0:-1], np.sqrt(cumul_err_heur2),'red', ax_hee.set_title("Heuristic disparity error")
bin_vals[0:-1], np.sqrt(cumul_err_nn2),'green', cross_out(plt, cross_out_mask)
bin_vals[0:-1], np.sqrt(cumul_diff_log2),'blue') plt.imshow(data[...,qsf.HEUR_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL, cmap=cmap_diff)
figs.append(fig) setlimsxy(lim_xy)
ffiles.append('cumulative') cross_out(plt, cross_out_mask)
ax_conf=plt.subplot(212) plt.colorbar(orientation='vertical') # location='bottom')
ax_conf.set_title("weights vs max9-min9")
plt.plot(bin_vals[0:-1], cumul_weights,'black') ax_nne=plt.subplot(326)
""" ax_nne.set_title("Network disparity error")
figs.append(fig) plt.imshow(data[...,qsf.NN_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL, cmap=cmap_diff)
ffiles.append('cumulative') setlimsxy(lim_xy)
pass cross_out(plt, cross_out_mask)
#bin_vals[0:-1] plt.colorbar(orientation='vertical') # location='bottom')
# fig.suptitle("Groud truth confidence") plt.tight_layout(rect =[0,0,1,TIGHT_TOP], h_pad = TIGHT_HPAD, w_pad = TIGHT_WPAD)
figs.append(fig)
fb_noext = os.path.splitext(os.path.basename(img_file))[0]#
# if subindex > 0:
# if subindex < 10:
# fb_noext+="abcdefghi"[subindex-1]
# else:
# fb_noext+="-"+str(subindex)
ffiles.append(fb_noext)
pass
# #
#how to allow adjustment before applying tight_layout? #how to allow adjustment before applying tight_layout?
......
nn_eval_lwir_00.py 0 → 100644
View file @ 4b02b283
#!/usr/bin/env python3
__copyright__ = "Copyright 2018, Elphel, Inc."
__license__ = "GPL-3.0+"
__email__ = "andrey@elphel.com"
#from PIL import Image
import os
import sys
#import glob
#import numpy as np
import imagej_tiffwriter
import time
import matplotlib.pyplot as plt
from matplotlib.backends.backend_pdf import PdfPages
import qcstereo_functions as qsf
import numpy as np
#import xml.etree.ElementTree as ET
qsf.TIME_START = time.time()
qsf.TIME_LAST = qsf.TIME_START
IMG_WIDTH = 20 # 324 # tiles per image row Defined in config
DEBUG_LEVEL= 1
try:
conf_file = sys.argv[1]
except IndexError:
print("Configuration path is required as a first argument. Optional second argument specifies root directory for data files")
exit(1)
try:
root_dir = sys.argv[2]
except IndexError:
root_dir = os.path.dirname(conf_file)
try:
modes = [sys.argv[3]] # train, infer
except IndexError:
modes = ['train']
print ("Configuration file: " + conf_file)
parameters, dirs, files, dbg_parameters = qsf.parseXmlConfig(conf_file, root_dir)
"""
Temporarily for backward compatibility
"""
if not "SLOSS_CLIP" in parameters:
parameters['SLOSS_CLIP'] = 0.5
print ("Old config, setting SLOSS_CLIP=", parameters['SLOSS_CLIP'])
"""
Defined in config file
"""
TILE_SIDE, TILE_LAYERS, TWO_TRAINS, NET_ARCH1, NET_ARCH2 = [None]*5
ABSOLUTE_DISPARITY,SYM8_SUB, WLOSS_LAMBDA, SLOSS_LAMBDA, SLOSS_CLIP = [None]*5
SPREAD_CONVERGENCE, INTER_CONVERGENCE, HOR_FLIP, DISP_DIFF_CAP, DISP_DIFF_SLOPE = [None]*5
CLUSTER_RADIUS,ABSOLUTE_DISPARITY = [None]*2
FGBG_MODE = 1 # 0 - do not filter by single-plane, 1 - remove split plabnes tiles, 2 - remove split planes and neighbors
FIGS_EXTENSIONS = ['png','pdf','svg']
#FIGS_ESXTENSIONS = ['png','pdf','svg']
EVAL_MODES = ["train","infer"]
FIGS_SAVESHOW = ['save','show']
globals().update(parameters)
try:
FIGS_EXTENSIONS = globals()['FIGS_ESXTENSIONS'] # fixing typo in configs
except:
pass
#exit(0)
TILE_SIZE = TILE_SIDE* TILE_SIDE # == 81
FEATURES_PER_TILE = TILE_LAYERS * TILE_SIZE# == 324
BATCH_SIZE = ([1,2][TWO_TRAINS])*2*1000//25 # == 80 Each batch of tiles has balanced D/S tiles, shuffled batches but not inside batches
SUFFIX=(str(NET_ARCH1)+'-'+str(NET_ARCH2)+
(["R","A"][ABSOLUTE_DISPARITY]) +
(["NS","S8"][SYM8_SUB])+
"WLAM"+str(WLOSS_LAMBDA)+
"SLAM"+str(SLOSS_LAMBDA)+
"SCLP"+str(SLOSS_CLIP)+
(['_nG','_G'][SPREAD_CONVERGENCE])+
(['_nI','_I'][INTER_CONVERGENCE]) +
(['_nHF',"_HF"][HOR_FLIP]) +
('_CP'+str(DISP_DIFF_CAP)) +
('_S'+str(DISP_DIFF_SLOPE))
)
##############################################################################
cluster_size = (2 * CLUSTER_RADIUS + 1) * (2 * CLUSTER_RADIUS + 1)
center_tile_index = 2 * CLUSTER_RADIUS * (CLUSTER_RADIUS + 1)
qsf.prepareFiles(dirs, files, suffix = SUFFIX)
#import tensorflow.contrib.slim as slim
#NN_DISP = 0
#HEUR_DISP = 1
#GT_DISP = 2
#GT_CONF = 3
#NN_NAN = 4
#HEUR_NAN = 5
#NN_DIFF = 6
#HEUR_DIFF = 7
# Now - more layers
CONF_MAX = 0.7
ERR_AMPL = 0.3
TIGHT_TOP = 0.95
TIGHT_HPAD = 1.0
TIGHT_WPAD = 1.0
FIGSIZE = [8.5,11.0]
WOI_COLOR = "red"
X_COLOR = "grey"
X_NEIBS = False
TRANSPARENT = True # for export
#dbg_parameters
def get_fig_params(disparity_ranges):
fig_params = []
for dr in disparity_ranges:
if dr[-1][0]=='-':
fig_params.append(None)
else:
subs = []
for s in dr[:-1]:
mm = s[:2]
try:
lims = s[2]
except IndexError:
lims = None
subs.append({'lim_val':mm, 'lim_xy':lims})
fig_params.append({'name':dr[-1],'ranges':subs})
return fig_params
#try:
fig_params = get_fig_params(dbg_parameters['disparity_ranges'])
pass
#temporary:
TIFF_ONLY = False # True
#max_bad = 2.5 # excludes only direct bad
max_bad = 2.5 #2.5 # 1.5 # excludes only direct bad
max_diff = 1.5 # 2.0 # 5.0 # maximal max-min difference
max_target_err = 1.0 # 0.5 # maximal max-min difference
max_disp = 5.0
min_strength = 0.18 #ignore tiles below
min_neibs = 1
max_log_to_mm = 0.5 # difference between center average and center should be under this fraction of max-min (0 - disables feature)
#num_bins = 256 # number of histogram bins
num_bins = 15 # 50 # number of histogram bins
use_gt_weights = True # False # True
index_gt = 2
index_gt_weight = 3
index_heur_err = 7
index_nn_err = 6
index_fgbg_sngl = 10
index_fgbg_neib = 11
index_mm = 23 # 8 # max-min
index_log = 24 # 9
index_bad = 25 # 10
index_num_neibs = 26 # 11
index_fgbg = [index_fgbg_sngl,index_fgbg_neib][X_NEIBS]
"""
Debugging high 9-tile variations, removing error for all tiles with lower difference between max and min
"""
#min_diff = 0.25 # remove all flat tiles with spread less than this (do not show on heuristic/network disparity errors subplots
min_diff = 0 # remove all flat tiles with spread less than this
max_target_err2 = max_target_err * max_target_err
if not 'show' in FIGS_SAVESHOW:
plt.ioff()
#for mode in ['train','infer']:
#for mode in ['infer']:
def cross_out(
plt,
cross_out_mask):
height = cross_out_mask.shape[0]
width = cross_out_mask.shape[1]
for row in range (height):
for col in range(width):
if cross_out_mask[row,col]:
xdata = [col-0.3, col+0.3]
ydata = [row-0.3, row+0.3]
plt.plot(xdata,ydata,color=X_COLOR)
ydata = [row+0.3, row-0.3]
plt.plot(xdata,ydata,color=X_COLOR)
for mode in modes: # ['train']:
figs = []
ffiles = [] # no ext
def setlimsxy(lim_xy):
if not lim_xy is None:
plt.xlim(min(lim_xy[:2]),max(lim_xy[:2]))
plt.ylim(max(lim_xy[2:]),min(lim_xy[2:]))
cumul_weights = None
for nfile, fpars in enumerate(fig_params):
if not fpars is None:
img_file = files['result'][nfile]
if mode == 'infer':
img_file = img_file.replace('.npy','-infer.npy')
"""
try:
# data,_ = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=True)
# data,_ = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
data,labels = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
except:
print ("Image file does not exist:", img_file)
continue
"""
pass
data,labels = qsf.result_npy_prepare(img_file, ABSOLUTE_DISPARITY, fix_nan=True, insert_deltas=3)
if True: #TIFF_ONLY:
tiff_path = img_file.replace('.npy','-test.tiff')
data = data.transpose(2,0,1)
print("Saving results to TIFF: "+tiff_path)
imagej_tiffwriter.save(tiff_path,data,labels=labels)
"""
Calculate histograms
"""
err_heur2 = data[index_heur_err]*data[index_heur_err]
err_nn2 = data[index_nn_err]* data[index_nn_err]
diff_log2 = data[index_log]* data[index_log]
weights = (
(data[index_gt] < max_disp) &
(err_heur2 < max_target_err2) &
(data[index_bad] < max_bad) &
(data[index_gt_weight] >= min_strength) &
(data[index_num_neibs] >= min_neibs)&
#max_log_to_mm = 0.5 # difference between center average and center should be under this fraction of max-min (0 - disables feature)
(data[index_log] < max_log_to_mm * np.sqrt(data[index_mm]) )
).astype(data.dtype) # 0.0/1.1
#max_disp
#max_target_err
if use_gt_weights:
weights *= data[index_gt_weight]
mm = data[index_mm]
weh = np.nan_to_num(weights*err_heur2)
wen = np.nan_to_num(weights*err_nn2)
wel = np.nan_to_num(weights*diff_log2)
hist_weights,bin_vals = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = weights, density = False)
hist_err_heur2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = weh, density = False)
hist_err_nn2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = wen, density = False)
hist_diff_log2,_ = np.histogram(a=mm, bins = num_bins, range = (0.0, max_diff), weights = wel, density = False)
if cumul_weights is None:
cumul_weights = hist_weights
cumul_err_heur2 = hist_err_heur2
cumul_err_nn2 = hist_err_nn2
cumul_diff_log2 = hist_diff_log2
else:
cumul_weights += hist_weights
cumul_err_heur2 += hist_err_heur2
cumul_err_nn2 += hist_err_nn2
cumul_diff_log2 += hist_diff_log2
hist_err_heur2 = np.nan_to_num(hist_err_heur2/hist_weights)
hist_err_nn2 = np.nan_to_num(hist_err_nn2/hist_weights)
hist_gain2 = np.nan_to_num(hist_err_heur2/hist_err_nn2)
hist_gain = np.sqrt(hist_gain2)
hist_diff_log2 = np.nan_to_num(hist_diff_log2/hist_weights)
print("hist_err_heur2", end = " ")
print(np.sqrt(hist_err_heur2))
print("hist_err_nn2", end = " ")
print(np.sqrt(hist_err_nn2))
print("hist_gain", end = " ")
print(hist_gain)
print("hist_diff_log2", end = " ")
print(np.sqrt(hist_diff_log2))
if min_diff> 0.0:
pass
good = (mm > min_diff).astype(mm.dtype)
good /= good # good - 1, bad - nan
data[index_heur_err] *= good
data[index_nn_err] *= good
data = data.transpose(1,2,0)
if TIFF_ONLY:
continue
cross_out_mask = data[...,index_fgbg] < 0.5 #data.shape = (15,20,27)
for subindex, rng in enumerate(fpars['ranges']):
lim_val = rng['lim_val']
lim_xy = rng['lim_xy']
fig = plt.figure(figsize=FIGSIZE)
fig.canvas.set_window_title(fpars['name'])
fig.suptitle(fpars['name'])
ax_conf=plt.subplot(322)
ax_conf.set_title("Ground truth confidence")
# fig.suptitle("Groud truth confidence")
plt.imshow(data[...,qsf.GT_CONF], vmin=0, vmax=CONF_MAX, cmap='gray')
if not lim_xy is None:
pass # show frame
xdata=[min(lim_xy[:2]),max(lim_xy[:2]),max(lim_xy[:2]),min(lim_xy[:2]),min(lim_xy[:2])]
ydata=[min(lim_xy[2:]),min(lim_xy[2:]),max(lim_xy[2:]),max(lim_xy[2:]),min(lim_xy[2:])]
plt.plot(xdata,ydata,color=WOI_COLOR)
# setlimsxy(lim_xy)
plt.colorbar(orientation='vertical') # location='bottom')
ax_gtd=plt.subplot(321)
ax_gtd.set_title("Ground truth disparity map")
plt.imshow(data[...,qsf.GT_DISP], vmin=lim_val[0], vmax=lim_val[1])
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
ax_hed=plt.subplot(323)
ax_hed.set_title("Heuristic disparity map")
plt.imshow(data[...,qsf.HEUR_NAN], vmin=lim_val[0], vmax=lim_val[1])
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
ax_nnd=plt.subplot(325)
ax_nnd.set_title("Network disparity output")
plt.imshow(data[...,qsf.NN_NAN], vmin=lim_val[0], vmax=lim_val[1])
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
ax_hee=plt.subplot(324)
ax_hee.set_title("Heuristic disparity error")
cross_out(plt, cross_out_mask)
plt.imshow(data[...,qsf.HEUR_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL)
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
ax_nne=plt.subplot(326)
ax_nne.set_title("Network disparity error")
plt.imshow(data[...,qsf.NN_DIFF], vmin=-ERR_AMPL, vmax=ERR_AMPL)
setlimsxy(lim_xy)
cross_out(plt, cross_out_mask)
plt.colorbar(orientation='vertical') # location='bottom')
plt.tight_layout(rect =[0,0,1,TIGHT_TOP], h_pad = TIGHT_HPAD, w_pad = TIGHT_WPAD)
figs.append(fig)
fb_noext = os.path.splitext(os.path.basename(img_file))[0]#
if subindex > 0:
if subindex < 10:
fb_noext+="abcdefghi"[subindex-1]
else:
fb_noext+="-"+str(subindex)
ffiles.append(fb_noext)
pass
if False: # True:
cumul_err_heur2 = np.nan_to_num(cumul_err_heur2/cumul_weights)
cumul_err_nn2 = np.nan_to_num(cumul_err_nn2/cumul_weights)
cumul_gain2 = np.nan_to_num(cumul_err_heur2/cumul_err_nn2)
cumul_gain = np.sqrt(cumul_gain2)
cumul_diff_log2 = np.nan_to_num(cumul_diff_log2/cumul_weights)
print("cumul_weights", end = " ")
print(cumul_weights)
print("cumul_err_heur", end = " ")
print(np.sqrt(cumul_err_heur2))
print("cumul_err_nn", end = " ")
print(np.sqrt(cumul_err_nn2))
print("cumul_gain", end = " ")
print(cumul_gain)
print("cumul_diff_log2", end = " ")
print(np.sqrt(cumul_diff_log2))
fig, ax1 = plt.subplots()
ax1.set_xlabel('3x3 tiles ground truth disparity max-min (pix)')
ax1.set_ylabel('RMSE\n(pix)', color='black', rotation='horizontal')
ax1.yaxis.set_label_coords(-0.045,0.92)
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_err_nn2), 'tab:red',label="network disparity RMSE")
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_err_heur2), 'tab:green',label="heuristic disparity RMSE")
ax1.plot(bin_vals[0:-1], np.sqrt(cumul_diff_log2), 'tab:cyan',label="ground truth LoG")
ax1.tick_params(axis='y', labelcolor='black')
ax2 = ax1.twinx() # instantiate a second axes that shares the same x-axis
ax2.set_ylabel('weight', color='black', rotation='horizontal') # we already handled the x-label with ax1
ax2.yaxis.set_label_coords(1.06,1.0)
ax2.plot(bin_vals[0:-1], cumul_weights,color='grey',dashes=[6, 2],label='weights = n_tiles * gt_confidence')
ax1.legend(loc="upper left", bbox_to_anchor=(0.2,1.0))
ax2.legend(loc="lower right", bbox_to_anchor=(1.0,0.1))
"""
fig = plt.figure(figsize=FIGSIZE)
fig.canvas.set_window_title('Cumulative')
fig.suptitle('Difference to GT')
# ax_conf=plt.subplot(322)
ax_conf=plt.subplot(211)
ax_conf.set_title("RMS vs max9-min9")
plt.plot(bin_vals[0:-1], np.sqrt(cumul_err_heur2),'red',
bin_vals[0:-1], np.sqrt(cumul_err_nn2),'green',
bin_vals[0:-1], np.sqrt(cumul_diff_log2),'blue')
figs.append(fig)
ffiles.append('cumulative')
ax_conf=plt.subplot(212)
ax_conf.set_title("weights vs max9-min9")
plt.plot(bin_vals[0:-1], cumul_weights,'black')
"""
figs.append(fig)
ffiles.append('cumulative')
pass
#bin_vals[0:-1]
# fig.suptitle("Groud truth confidence")
#
#how to allow adjustment before applying tight_layout?
pass
for fig in figs:
fig.tight_layout(rect =[0,0,1,TIGHT_TOP], h_pad = TIGHT_HPAD, w_pad = TIGHT_WPAD)
if FIGS_EXTENSIONS and figs and 'save' in FIGS_SAVESHOW:
try:
print ("Creating output directory for figures: ",dirs['figures'])
os.makedirs(dirs['figures'])
except:
pass
pp=None
if 'pdf' in FIGS_EXTENSIONS:
if mode == 'infer':
pdf_path = os.path.join(dirs['figures'],"figures-infer%s.pdf"%str(min_diff))
else:
pdf_path = os.path.join(dirs['figures'],"figures-train%s.pdf"%str(min_diff))
pp= PdfPages(pdf_path)
for fb_noext, fig in zip(ffiles,figs):
for ext in FIGS_EXTENSIONS:
if ext == 'pdf':
pass
fig.savefig(pp,format='pdf')
else:
if mode == 'infer':
noext = fb_noext+'-infer'
else:
noext = fb_noext+'-train'
fig.savefig(
fname = os.path.join(dirs['figures'],noext+"."+ext),
transparent = TRANSPARENT,
)
pass
if pp:
pp.close()
if 'show' in FIGS_SAVESHOW:
plt.show()
#FIGS_ESXTENSIONS
#qsf.evaluateAllResults(result_files = files['result'],
# absolute_disparity = ABSOLUTE_DISPARITY,
# cluster_radius = CLUSTER_RADIUS)
print("All done")
exit (0)
qcstereo_network.py
View file @ 4b02b283
...@@ -110,21 +110,21 @@ def network_sub(input_tensor, ...@@ -110,21 +110,21 @@ def network_sub(input_tensor,
fc_sym.append(slim.fully_connected(inp8[j], num_sym8, activation_fn=lrelu, scope= scp, reuse = reuse_this)) fc_sym.append(slim.fully_connected(inp8[j], num_sym8, activation_fn=lrelu, scope= scp, reuse = reuse_this))
if not reuse_this: if not reuse_this:
with tf.compat.v1.variable_scope(scp,reuse=True) : # tf.AUTO_REUSE): with tf.compat.v1.variable_scope(scp,reuse=True) : # tf.AUTO_REUSE):
inp_weights.append(tf.get_variable('weights')) # ,shape=[inp.shape[1],num_outs])) inp_weights.append(tf.compat.v1.get_variable('weights')) # ,shape=[inp.shape[1],num_outs]))
if num_non_sum > 0: if num_non_sum > 0:
reuse_this = reuse reuse_this = reuse
scp = 'g_fc_sub'+str(i)+"r" scp = 'g_fc_sub'+str(i)+"r"
fc_sym.append(slim.fully_connected(inp, num_non_sum, activation_fn=lrelu, scope=scp, reuse = reuse_this)) fc_sym.append(slim.fully_connected(inp, num_non_sum, activation_fn=lrelu, scope=scp, reuse = reuse_this))
if not reuse_this: if not reuse_this:
with tf.compat.v1.variable_scope(scp,reuse=True) : # tf.AUTO_REUSE): with tf.compat.v1.variable_scope(scp,reuse=True) : # tf.AUTO_REUSE):
inp_weights.append(tf.get_variable('weights')) # ,shape=[inp.shape[1],num_outs])) inp_weights.append(tf.compat.v1.get_variable('weights')) # ,shape=[inp.shape[1],num_outs]))
fc.append(tf.concat(fc_sym, 1, name='sym_input_layer')) fc.append(tf.concat(fc_sym, 1, name='sym_input_layer'))
else: else:
scp = 'g_fc_sub'+str(i) scp = 'g_fc_sub'+str(i)
fc.append(slim.fully_connected(inp, num_outs, activation_fn=lrelu, scope= scp, reuse = reuse)) fc.append(slim.fully_connected(inp, num_outs, activation_fn=lrelu, scope= scp, reuse = reuse))
if not reuse: if not reuse:
with tf.compat.v1.variable_scope(scp, reuse=True) : # tf.AUTO_REUSE): with tf.compat.v1.variable_scope(scp, reuse=True) : # tf.AUTO_REUSE):
inp_weights.append(tf.get_variable('weights')) # ,shape=[inp.shape[1],num_outs])) inp_weights.append(tf.compat.v1.get_variable('weights')) # ,shape=[inp.shape[1],num_outs]))
return fc[-1], inp_weights return fc[-1], inp_weights
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment