Skip to content

P
python3-imagej-tiff
Commit d7fc8ac4 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

Modyfying costs, feeding datasets to placeholders instead of constants

parent 47ad3d9f
Show whitespace changes
Inline Side-by-side
Showing with 1604 additions and 55 deletions
explore_data.py
View file @ d7fc8ac4
...@@ -225,12 +225,63 @@ class ExploreData: ...@@ -225,12 +225,63 @@ class ExploreData:
self.good_tiles = self.blurred_hist >= h_cutoff self.good_tiles = self.blurred_hist >= h_cutoff
self.blurred_hist *= self.good_tiles # set bad ones to zero self.blurred_hist *= self.good_tiles # set bad ones to zero
def exploreNeibs(self,
data_ds, # disparity/strength data for all files (train or test)
radius, # how far to look from center each side ( 1- 3x3, 2 - 5x5)
disp_thesh = 5.0): # reduce effective variance for higher disparities
"""
For each tile calculate difference between max and min among neighbors and number of qualifying neighbors (bad cewnter is not removed)
"""
disp_min = np.empty_like(data_ds[...,0], dtype = np.float)
disp_max = np.empty_like(disp_min, dtype = np.float)
tile_neibs = np.zeros_like(disp_min, dtype = np.int)
dmin = data_ds[...,0].min()
dmax = data_ds[...,0].max()
good_tiles = self.getBB(data_ds) >= 0
side = 2 * radius + 1
for nf, ds in enumerate(data_ds):
disp = ds[...,0]
height = disp.shape[0]
width = disp.shape[1]
bad_max = np.ones((height+side, width+side), dtype=float) * dmax
bad_min = np.ones((height+side, width+side), dtype=float) * dmin
good = np.zeros((height+side, width+side), dtype=int)
#Assign centers of the array, replace bad tiles with max/min (so they will not change min/max)
bad_max[radius:height+radius,radius:width+radius] = np.select([good_tiles[nf]],[disp],default = dmax)
bad_min[radius:height+radius,radius:width+radius] = np.select([good_tiles[nf]],[disp],default = dmin)
good [radius:height+radius,radius:width+radius] = good_tiles[nf]
disp_min [nf,...] = disp
disp_max [nf,...] = disp
tile_neibs[nf,...] = good_tiles[nf]
for offset_y in range(-radius, radius+1):
oy = offset_y+radius
for offset_x in range(-radius, radius+1):
ox = offset_x+radius
if offset_y or offset_x: # Skip center - already copied
np.minimum(disp_min[nf], bad_max[oy:oy+height, ox:ox+width], out=disp_min[nf])
np.maximum(disp_max[nf], bad_min[oy:oy+height, ox:ox+width], out=disp_max[nf])
tile_neibs[nf] += good[oy:oy+height, ox:ox+width]
pass
pass
pass
pass
#disp_thesh
disp_avar = disp_max - disp_min
disp_rvar = disp_avar * disp_thesh / disp_max
disp_var = np.select([disp_max >= disp_thesh, disp_max < disp_thesh],[disp_rvar,disp_avar])
return disp_var, tile_neibs
def assignBatchBins(self, def assignBatchBins(self,
disp_bins, disp_bins,
str_bins, str_bins,
files_per_scene = 5, # not used here, will be used when generating batches files_per_scene = 5, # not used here, will be used when generating batches
min_batch_choices=10, # not used here, will be used when generating batches min_batch_choices=10, # not used here, will be used when generating batches
max_batch_files = 10): # not used here, will be used when generating batches max_batch_files = 10): # not used here, will be used when generating batches
"""
for each disparity/strength combination (self.disparity_bins * self.strength_bins = 1000*100) provide number of "large"
variable-size disparity/strength bin, or -1 if this disparity/strength combination does not seem right
"""
self.files_per_scene = files_per_scene self.files_per_scene = files_per_scene
self.min_batch_choices=min_batch_choices self.min_batch_choices=min_batch_choices
self.max_batch_files = max_batch_files self.max_batch_files = max_batch_files
...@@ -244,8 +295,8 @@ class ExploreData: ...@@ -244,8 +295,8 @@ class ExploreData:
disp_run_tot = 0.0 disp_run_tot = 0.0
disp_batch = 0 disp_batch = 0
disp=0 disp=0
disp_hist = np.linspace(0,disp_bins * str_bins,disp_bins+1)
num_batch_bins = disp_bins * str_bins num_batch_bins = disp_bins * str_bins
disp_hist = np.linspace(0, num_batch_bins, disp_bins+1)
batch_index = 0 batch_index = 0
num_members = np.zeros((num_batch_bins,),int) num_members = np.zeros((num_batch_bins,),int)
while disp_batch < disp_bins: while disp_batch < disp_bins:
...@@ -293,29 +344,52 @@ class ExploreData: ...@@ -293,29 +344,52 @@ class ExploreData:
self.hist_to_batch = hist_to_batch self.hist_to_batch = hist_to_batch
return hist_to_batch return hist_to_batch
def makeBatchLists(self, def getBB(self, data_ds):
train_ds = None): """
if train_ds is None: for each file, each tile get histogram index (or -1 for bad tiles)
train_ds = self.train_ds """
hist_to_batch = self.hist_to_batch hist_to_batch = self.hist_to_batch
files_batch_list = [] files_batch_list = []
disp_step = ( self.disparity_max_clip - self.disparity_min_clip )/ self.disparity_bins disp_step = ( self.disparity_max_clip - self.disparity_min_clip )/ self.disparity_bins
str_step = ( self.strength_max_clip - self.strength_min_clip )/ self.strength_bins str_step = ( self.strength_max_clip - self.strength_min_clip )/ self.strength_bins
bb = np.empty((train_ds.shape[0],train_ds.shape[1],train_ds.shape[2]),int) bb = np.empty_like(data_ds[...,0],dtype=int)
num_batch_tiles = np.empty((train_ds.shape[0],self.hist_to_batch.max()+1),dtype = int) for findx in range(data_ds.shape[0]):
for findx in range(train_ds.shape[0]): ds = data_ds[findx]
ds = train_ds[findx] gt = ds[...,1] > 0.0 # all true - check
db = (((ds[...,0] - self.disparity_min_clip)/disp_step).astype(int))*gt
sb = (((ds[...,1] - self.strength_min_clip)/ str_step).astype(int))*gt
np.clip(db, 0, self.disparity_bins-1, out = db)
np.clip(sb, 0, self.strength_bins-1, out = sb)
bb[findx] = (self.hist_to_batch[sb.reshape(self.num_tiles),db.reshape(self.num_tiles)]) .reshape(db.shape[0],db.shape[1]) + (gt -1)
return bb
def makeBatchLists(self,
data_ds = None, # (disparity,strength) per scene, per tile
disp_var = None, # difference between maximal and minimal disparity for each scene, each tile
disp_neibs = None, # number of valid tiles around each center tile (for 3x3 (radius = 1) - macximal is 9
min_var = None, # Minimal tile variance to include
max_var = None, # Maximal tile variance to include
min_neibs = None):# Minimal number of valid tiles to include
if data_ds is None:
data_ds = self.train_ds
hist_to_batch = self.hist_to_batch
num_batch_tiles = np.empty((data_ds.shape[0],self.hist_to_batch.max()+1),dtype = int)
bb = self.getBB(data_ds)
use_neibs = not ((disp_var is None) or (disp_neibs is None) or (min_var is None) or (max_var is None) or (min_neibs is None))
'''
bb = np.empty((data_ds.shape[0],data_ds.shape[1],data_ds.shape[2]),int)
for findx in range(data_ds.shape[0]):
ds = data_ds[findx]
gt = ds[...,1] > 0.0 # all true - check gt = ds[...,1] > 0.0 # all true - check
db = (((ds[...,0] - self.disparity_min_clip)/disp_step).astype(int))*gt db = (((ds[...,0] - self.disparity_min_clip)/disp_step).astype(int))*gt
sb = (((ds[...,1] - self.strength_min_clip)/ str_step).astype(int))*gt sb = (((ds[...,1] - self.strength_min_clip)/ str_step).astype(int))*gt
np.clip(db, 0, self.disparity_bins-1, out = db) np.clip(db, 0, self.disparity_bins-1, out = db)
np.clip(sb, 0, self.strength_bins-1, out = sb) np.clip(sb, 0, self.strength_bins-1, out = sb)
bb[findx] = (self.hist_to_batch[sb.reshape(self.num_tiles),db.reshape(self.num_tiles)]).reshape(db.shape[0],db.shape[1]) + (gt -1) bb[findx] = (self.hist_to_batch[sb.reshape(self.num_tiles),db.reshape(self.num_tiles)]) .reshape(db.shape[0],db.shape[1]) + (gt -1)
pass pass
# return bb '''
list_of_file_lists=[] list_of_file_lists=[]
for findx in range(train_ds.shape[0]): for findx in range(data_ds.shape[0]):
foffs = findx * self.num_tiles foffs = findx * self.num_tiles
lst = [] lst = []
for i in range (self.hist_to_batch.max()+1): for i in range (self.hist_to_batch.max()+1):
...@@ -323,6 +397,15 @@ class ExploreData: ...@@ -323,6 +397,15 @@ class ExploreData:
# bb1d = bb[findx].reshape(self.num_tiles) # bb1d = bb[findx].reshape(self.num_tiles)
for n, indx in enumerate(bb[findx].reshape(self.num_tiles)): for n, indx in enumerate(bb[findx].reshape(self.num_tiles)):
if indx >= 0: if indx >= 0:
if use_neibs:
disp_var_tiles = disp_var[findx].reshape(self.num_tiles)
disp_neibs_tiles = disp_neibs[findx].reshape(self.num_tiles)
if disp_neibs_tiles[indx] < min_neibs:
continue # too few neighbors
if not disp_var_tiles[indx] >= min_var:
continue #too small variance
if not disp_var_tiles[indx] < max_var:
continue #too large variance
lst[indx].append(foffs + n) lst[indx].append(foffs + n)
lst_arr=[] lst_arr=[]
for i,l in enumerate(lst): for i,l in enumerate(lst):
...@@ -503,9 +586,9 @@ class ExploreData: ...@@ -503,9 +586,9 @@ class ExploreData:
dtype_target_disparity = _dtype_feature(target_disparity_batch_shuffled) dtype_target_disparity = _dtype_feature(target_disparity_batch_shuffled)
dtype_feature_gt_ds = _dtype_feature(gt_ds_batch_shuffled) dtype_feature_gt_ds = _dtype_feature(gt_ds_batch_shuffled)
for i in range(tiles_in_batch): for i in range(tiles_in_batch):
x = corr2d_batch_shuffled[i] x = corr2d_batch_shuffled[i].astype(np.float32)
y = target_disparity_batch_shuffled[i] y = target_disparity_batch_shuffled[i].astype(np.float32)
z = gt_ds_batch_shuffled[i] z = gt_ds_batch_shuffled[i].astype(np.float32)
d_feature = {'corr2d': dtype_feature_corr2d(x), d_feature = {'corr2d': dtype_feature_corr2d(x),
'target_disparity':dtype_target_disparity(y), 'target_disparity':dtype_target_disparity(y),
'gt_ds': dtype_feature_gt_ds(z)} 'gt_ds': dtype_feature_gt_ds(z)}
...@@ -516,6 +599,70 @@ class ExploreData: ...@@ -516,6 +599,70 @@ class ExploreData:
writer.close() writer.close()
sys.stdout.flush() sys.stdout.flush()
def showVariance(self,
rds_list, # list of disparity/strength files, suchas training, testing
disp_var_list, # list of disparity variance files. Same shape(but last dim) as rds_list
num_neibs_list, # list of number of tile neibs files. Same shape(but last dim) as rds_list
variance_min = 0.0,
variance_max = 1.5,
neibs_min = 9,
#Same parameters as for the histogram
# disparity_bins = 1000,
# strength_bins = 100,
# disparity_min_drop = -0.1,
# disparity_min_clip = -0.1,
# disparity_max_drop = 100.0,
# disparity_max_clip = 100.0,
# strength_min_drop = 0.1,
# strength_min_clip = 0.1,
# strength_max_drop = 1.0,
# strength_max_clip = 0.9,
normalize = False): # True):
good_tiles_list=[]
for nf, combo_rds in enumerate(rds_list):
disp_var = disp_var_list[nf]
num_neibs = num_neibs_list[nf]
good_tiles = np.empty((combo_rds.shape[0], combo_rds.shape[1],combo_rds.shape[2]), dtype=bool)
for ids in range (combo_rds.shape[0]): #iterate over all scenes ds[2][rows][cols]
ds = combo_rds[ids]
disparity = ds[...,0]
strength = ds[...,1]
variance = disp_var[ids]
neibs = num_neibs[ids]
good_tiles[ids] = disparity >= self.disparity_min_drop
good_tiles[ids] &= disparity <= self.disparity_max_drop
good_tiles[ids] &= strength >= self.strength_min_drop
good_tiles[ids] &= strength <= self.strength_max_drop
good_tiles[ids] &= neibs >= neibs_min
good_tiles[ids] &= variance >= variance_min
good_tiles[ids] &= variance < variance_max
disparity = np.nan_to_num(disparity, copy = False) # to be able to multiply by 0.0 in mask | copy=False, then out=disparity all done in-place
strength = np.nan_to_num(strength, copy = False) # likely should never happen
np.clip(disparity, self.disparity_min_clip, self.disparity_max_clip, out = disparity)
np.clip(strength, self.strength_min_clip, self.strength_max_clip, out = strength)
good_tiles_list.append(good_tiles)
combo_rds = np.concatenate(rds_list)
hist, xedges, yedges = np.histogram2d( # xedges, yedges - just for debugging
x = combo_rds[...,1].flatten(),
y = combo_rds[...,0].flatten(),
bins= (self.strength_bins, self.disparity_bins),
range= ((self.strength_min_clip,self.strength_max_clip),(self.disparity_min_clip,self.disparity_max_clip)),
normed= normalize,
weights= np.concatenate(good_tiles_list).flatten())
mytitle = "Disparity_Strength variance histogram"
fig = plt.figure()
fig.canvas.set_window_title(mytitle)
fig.suptitle("Min variance = %f, max variance = %f, min neibs = %d"%(variance_min, variance_max, neibs_min))
# plt.imshow(hist, vmin=0, vmax=.1 * hist.max())#,vmin=-6,vmax=-2) # , vmin=0, vmax=.01)
plt.imshow(hist, vmin=0.0, vmax=300.0)#,vmin=-6,vmax=-2) # , vmin=0, vmax=.01)
plt.colorbar(orientation='horizontal') # location='bottom')
# for i, combo_rds in enumerate(rds_list):
# for ids in range (combo_rds.shape[0]): #iterate over all scenes ds[2][rows][cols]
# combo_rds[ids][...,1]*= good_tiles_list[i][ids]
# return hist, xedges, yedges
#MAIN #MAIN
if __name__ == "__main__": if __name__ == "__main__":
try: try:
...@@ -530,13 +677,16 @@ if __name__ == "__main__": ...@@ -530,13 +677,16 @@ if __name__ == "__main__":
try: try:
train_filenameTFR = sys.argv[3] train_filenameTFR = sys.argv[3]
except IndexError: except IndexError:
train_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/train.tfrecords" train_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/train_01.tfrecords"
try: try:
test_filenameTFR = sys.argv[4] test_filenameTFR = sys.argv[4]
except IndexError: except IndexError:
test_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/test.tfrecords" test_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/test_01.tfrecords"
#Parameters to generate neighbors data. Set radius to 0 to generate single-tile
RADIUS = 1
MIN_NEIBS = (2 * RADIUS + 1) * (2 * RADIUS + 1) # All tiles valid
VARIANCE_THRESHOLD = 1.5
# corr2d, target_disparity, gt_ds = readTFRewcordsEpoch(train_filenameTFR) # corr2d, target_disparity, gt_ds = readTFRewcordsEpoch(train_filenameTFR)
# print_time("Read %d tiles"%(corr2d.shape[0])) # print_time("Read %d tiles"%(corr2d.shape[0]))
# exit (0) # exit (0)
...@@ -575,14 +725,43 @@ if __name__ == "__main__": ...@@ -575,14 +725,43 @@ if __name__ == "__main__":
plt.imshow(bb_display) #, vmin=0, vmax=.1 * ex_data.blurred_hist.max())#,vmin=-6,vmax=-2) # , vmin=0, vmax=.01) plt.imshow(bb_display) #, vmin=0, vmax=.1 * ex_data.blurred_hist.max())#,vmin=-6,vmax=-2) # , vmin=0, vmax=.01)
""" prepare test dataset """ """ prepare test dataset """
# RADIUS = 1
# MIN_NEIBS = (2 * RADIUS + 1) * (2 * RADIUS + 1) # All tiles valid
# VARIANCE_THRESHOLD = 1.5
if (RADIUS > 0):
disp_var_test, num_neibs_test = ex_data.exploreNeibs(ex_data.test_ds, RADIUS)
disp_var_train, num_neibs_train = ex_data.exploreNeibs(ex_data.train_ds, RADIUS)
for var_thresh in [0.1, 1.0, 1.5, 2.0, 5.0]:
ex_data.showVariance(
rds_list = [ex_data.train_ds, ex_data.test_ds], # list of disparity/strength files, suchas training, testing
disp_var_list = [disp_var_train, disp_var_test], # list of disparity variance files. Same shape(but last dim) as rds_list
num_neibs_list = [num_neibs_train, num_neibs_test], # list of number of tile neibs files. Same shape(but last dim) as rds_list
variance_min = 0.0,
variance_max = var_thresh,
neibs_min = 9)
ex_data.showVariance(
rds_list = [ex_data.train_ds, ex_data.test_ds], # list of disparity/strength files, suchas training, testing
disp_var_list = [disp_var_train, disp_var_test], # list of disparity variance files. Same shape(but last dim) as rds_list
num_neibs_list = [num_neibs_train, num_neibs_test], # list of number of tile neibs files. Same shape(but last dim) as rds_list
variance_min = var_thresh,
variance_max = 1000.0,
neibs_min = 9)
pass
pass
# show varinace histogram
else:
disp_var_test, num_neibs_test = None, None
disp_var_train, num_neibs_train = None, None
ml_list=ex_data.getMLList(ex_data.files_test) ml_list=ex_data.getMLList(ex_data.files_test)
ex_data.makeBatchLists(train_ds = ex_data.test_ds) ex_data.makeBatchLists(data_ds = ex_data.test_ds)
ex_data.writeTFRewcordsEpoch(test_filenameTFR, test_set=True) ex_data.writeTFRewcordsEpoch(test_filenameTFR, test_set=True)
""" prepare train dataset """ """ prepare train dataset """
ml_list=ex_data.getMLList(ex_data.files_train) # train_list) ml_list=ex_data.getMLList(ex_data.files_train) # train_list)
ex_data.makeBatchLists(train_ds = ex_data.train_ds) ex_data.makeBatchLists(data_ds = ex_data.train_ds)
ex_data.writeTFRewcordsEpoch(train_filenameTFR,test_set = False) ex_data.writeTFRewcordsEpoch(train_filenameTFR,test_set = False)
......
nn_ds_inmem2.py 0 → 100644
View file @ d7fc8ac4
#!/usr/bin/env python3
from numpy import float64
__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 itertools
import time
import matplotlib.pyplot as plt
import shutil
TIME_START = time.time()
TIME_LAST = TIME_START
DEBUG_LEVEL= 1
DISP_BATCH_BINS = 20 # Number of batch disparity bins
STR_BATCH_BINS = 10 # Number of batch strength bins
FILES_PER_SCENE = 5 # number of random offset files for the scene to select from (0 - use all available)
#MIN_BATCH_CHOICES = 10 # minimal number of tiles in a file for each bin to select from
#MAX_BATCH_FILES = 10 #maximal number of files to use in a batch
MAX_EPOCH = 500
LR = 1e-4 # learning rate
USE_CONFIDENCE = False
ABSOLUTE_DISPARITY = True # False
DEBUG_PLT_LOSS = True
FEATURES_PER_TILE = 324
EPOCHS_TO_RUN = 10000 #0
RUN_TOT_AVG = 100 # last batches to average. Epoch is 307 training batches
BATCH_SIZE = 1000 # Each batch of tiles has balanced D/S tiles, shuffled batches but not inside batches
SHUFFLE_EPOCH = True
#DEBUG_PACK_TILES = True
#http://stackoverflow.com/questions/287871/print-in-terminal-with-colors-using-python
class bcolors:
HEADER = '\033[95m'
OKBLUE = '\033[94m'
OKGREEN = '\033[92m'
WARNING = '\033[38;5;214m'
FAIL = '\033[91m'
ENDC = '\033[0m'
BOLD = '\033[1m'
BOLDWHITE = '\033[1;37m'
UNDERLINE = '\033[4m'
def print_time(txt="",end="\n"):
global TIME_LAST
t = time.time()
if txt:
txt +=" "
print(("%s"+bcolors.BOLDWHITE+"at %.4fs (+%.4fs)"+bcolors.ENDC)%(txt,t-TIME_START,t-TIME_LAST), end = end, flush=True)
TIME_LAST = t
#reading to memory (testing)
def readTFRewcordsEpoch(train_filename):
# filenames = [train_filename]
# dataset = tf.data.TFRecordDataset(filenames)
if not '.tfrecords' in train_filename:
train_filename += '.tfrecords'
record_iterator = tf.python_io.tf_record_iterator(path=train_filename)
corr2d_list=[]
target_disparity_list=[]
gt_ds_list = []
for string_record in record_iterator:
example = tf.train.Example()
example.ParseFromString(string_record)
corr2d_list.append (np.array(example.features.feature['corr2d'].float_list.value, dtype=np.float32))
# target_disparity_list.append(np.array(example.features.feature['target_disparity'].float_list.value[0], dtype=np.float32))
target_disparity_list.append (np.array(example.features.feature['target_disparity'].float_list.value, dtype=np.float32))
gt_ds_list.append (np.array(example.features.feature['gt_ds'].float_list.value, dtype= np.float32))
corr2d= np.array(corr2d_list)
target_disparity = np.array(target_disparity_list)
gt_ds = np.array(gt_ds_list)
return corr2d, target_disparity, gt_ds
#from http://warmspringwinds.github.io/tensorflow/tf-slim/2016/12/21/tfrecords-guide/
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'corr2d': tf.FixedLenFeature([324],tf.float32), #string),
'target_disparity': tf.FixedLenFeature([1], tf.float32), #.string),
'gt_ds': tf.FixedLenFeature([2], tf.float32) #.string)
})
corr2d = features['corr2d'] # tf.decode_raw(features['corr2d'], tf.float32)
target_disparity = features['target_disparity'] # tf.decode_raw(features['target_disparity'], tf.float32)
gt_ds = tf.cast(features['gt_ds'], tf.float32) # tf.decode_raw(features['gt_ds'], tf.float32)
in_features = tf.concat([corr2d,target_disparity],0)
# still some nan-s in correlation data?
# in_features_clean = tf.where(tf.is_nan(in_features), tf.zeros_like(in_features), in_features)
# corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features_clean, target_disparity, gt_ds],
corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features, target_disparity, gt_ds],
batch_size=1000, # 2,
capacity=30,
num_threads=2,
min_after_dequeue=10)
return corr2d_out, target_disparity_out, gt_ds_out
#http://adventuresinmachinelearning.com/introduction-tensorflow-queuing/
#Main code
try:
train_filenameTFR = sys.argv[1]
except IndexError:
train_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/train.tfrecords"
try:
test_filenameTFR = sys.argv[2]
except IndexError:
test_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/test.tfrecords"
#FILES_PER_SCENE
#print_time("Importing TensorCrawl")
import tensorflow as tf
import tensorflow.contrib.slim as slim
#print_time("TensorCrawl imported")
print_time("Importing training data... ", end="")
corr2d_train, target_disparity_train, gt_ds_train = readTFRewcordsEpoch(train_filenameTFR)
print_time(" Done")
corr2d_train_placeholder = tf.placeholder(corr2d_train.dtype, (None,324)) # corr2d_train.shape)
target_disparity_train_placeholder = tf.placeholder(target_disparity_train.dtype, (None,1)) #target_disparity_train.shape)
gt_ds_train_placeholder = tf.placeholder(gt_ds_train.dtype, (None,2)) #gt_ds_train.shape)
dataset_train = tf.data.Dataset.from_tensor_slices({
"corr2d":corr2d_train_placeholder,
"target_disparity": target_disparity_train_placeholder,
"gt_ds": gt_ds_train_placeholder})
dataset_train_size = len(corr2d_train)
print_time("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
dataset_train = dataset_train.batch(BATCH_SIZE)
dataset_train_size //= BATCH_SIZE
print("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
iterator_train = dataset_train.make_initializable_iterator()
next_element_train = iterator_train.get_next()
print_time("Importing test data... ", end="")
corr2d_test, target_disparity_test, gt_ds_test = readTFRewcordsEpoch(test_filenameTFR)
print_time(" Done")
"""
dataset_test = tf.data.Dataset.from_tensor_slices({
"corr2d":corr2d_test,
"target_disparity": target_disparity_test,
"gt_ds": gt_ds_test})
"""
dataset_test_size = len(corr2d_test)
#print_time("dataset_test.output_types "+str(dataset_test.output_types)+", dataset_test.output_shapes "+str(dataset_test.output_shapes)+", number of elements="+str(dataset_test_size))
#dataset_test = dataset_test.batch(BATCH_SIZE)
dataset_test_size //= BATCH_SIZE
#print("dataset_test.output_types "+str(dataset_test.output_types)+", dataset_test.output_shapes "+str(dataset_test.output_shapes)+", number of elements="+str(dataset_test_size))
"""
iterator_test = dataset_test.make_initializable_iterator()
next_element_test = iterator_test.get_next()
"""
#https://www.tensorflow.org/versions/r1.5/programmers_guide/datasets
result_dir = './attic/result_inmem2/'
checkpoint_dir = './attic/result_inmem2/'
save_freq = 500
def lrelu(x):
return tf.maximum(x*0.2,x)
# return tf.nn.relu(x)
def network(input):
# fc1 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc1')
# fc2 = slim.fully_connected(fc1, 128, activation_fn=lrelu,scope='g_fc2')
## fc3 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc3')
## fc4 = slim.fully_connected(fc3, 128, activation_fn=lrelu,scope='g_fc4')
## fc5 = slim.fully_connected(fc4, 64, activation_fn=lrelu,scope='g_fc5')
fc3 = slim.fully_connected(input, 32, activation_fn=lrelu,scope='g_fc3')
fc4 = slim.fully_connected(fc3, 20, activation_fn=lrelu,scope='g_fc4')
fc5 = slim.fully_connected(fc4, 16, activation_fn=lrelu,scope='g_fc5')
if USE_CONFIDENCE:
fc6 = slim.fully_connected(fc5, 2, activation_fn=lrelu,scope='g_fc6')
else:
fc6 = slim.fully_connected(fc5, 1, activation_fn=None,scope='g_fc6')
#If using residual disparity, split last layer into 2 or remove activation and add rectifier to confidence only
return fc6
def batchLoss(out_batch, # [batch_size,(1..2)] tf_result
target_disparity_batch, # [batch_size] tf placeholder
gt_ds_batch, # [batch_size,2] tf placeholder
absolute_disparity = True, #when false there should be no activation on disparity output !
use_confidence = True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 1.0,
error2_offset = 0.0025): # (0.05^2)
with tf.name_scope("BatchLoss"):
"""
Here confidence should be after relU. Disparity - may be also if absolute, but no activation if output is residual disparity
"""
tf_lambda_conf_avg = tf.constant(lambda_conf_avg, dtype=tf.float32, name="tf_lambda_conf_avg")
tf_lambda_conf_pwr = tf.constant(lambda_conf_pwr, dtype=tf.float32, name="tf_lambda_conf_pwr")
tf_conf_pwr = tf.constant(conf_pwr, dtype=tf.float32, name="tf_conf_pwr")
tf_gt_conf_offset = tf.constant(gt_conf_offset, dtype=tf.float32, name="tf_gt_conf_offset")
tf_gt_conf_pwr = tf.constant(gt_conf_pwr, dtype=tf.float32, name="tf_gt_conf_pwr")
tf_num_tiles = tf.shape(gt_ds_batch)[0]
tf_0f = tf.constant(0.0, dtype=tf.float32, name="tf_0f")
tf_1f = tf.constant(1.0, dtype=tf.float32, name="tf_1f")
tf_maxw = tf.constant(1.0, dtype=tf.float32, name="tf_maxw")
if gt_conf_pwr == 0:
w = tf.ones((out_batch.shape[0]), dtype=tf.float32,name="w_ones")
else:
# w_slice = tf.slice(gt_ds_batch,[0,1],[-1,1], name = "w_gt_slice")
w_slice = tf.reshape(gt_ds_batch[:,1],[-1], name = "w_gt_slice")
w_sub = tf.subtract (w_slice, tf_gt_conf_offset, name = "w_sub")
# w_clip = tf.clip_by_value(w_sub, tf_0f,tf_maxw, name = "w_clip")
w_clip = tf.maximum(w_sub, tf_0f, name = "w_clip")
if gt_conf_pwr == 1.0:
w = w_clip
else:
w=tf.pow(w_clip, tf_gt_conf_pwr, name = "w_pow")
if use_confidence:
tf_num_tilesf = tf.cast(tf_num_tiles, dtype=tf.float32, name="tf_num_tilesf")
# conf_slice = tf.slice(out_batch,[0,1],[-1,1], name = "conf_slice")
conf_slice = tf.reshape(out_batch[:,1],[-1], name = "conf_slice")
conf_sum = tf.reduce_sum(conf_slice, name = "conf_sum")
conf_avg = tf.divide(conf_sum, tf_num_tilesf, name = "conf_avg")
conf_avg1 = tf.subtract(conf_avg, tf_1f, name = "conf_avg1")
conf_avg2 = tf.square(conf_avg1, name = "conf_avg2")
cost2 = tf.multiply (conf_avg2, tf_lambda_conf_avg, name = "cost2")
iconf_avg = tf.divide(tf_1f, conf_avg, name = "iconf_avg")
nconf = tf.multiply (conf_slice, iconf_avg, name = "nconf") #normalized confidence
nconf_pwr = tf.pow(nconf, conf_pwr, name = "nconf_pwr")
nconf_pwr_sum = tf.reduce_sum(nconf_pwr, name = "nconf_pwr_sum")
nconf_pwr_offs = tf.subtract(nconf_pwr_sum, tf_1f, name = "nconf_pwr_offs")
cost3 = tf.multiply (conf_avg2, nconf_pwr_offs, name = "cost3")
w_all = tf.multiply (w, nconf, name = "w_all")
else:
w_all = w
# cost2 = 0.0
# cost3 = 0.0
# normalize weights
w_sum = tf.reduce_sum(w_all, name = "w_sum")
iw_sum = tf.divide(tf_1f, w_sum, name = "iw_sum")
w_norm = tf.multiply (w_all, iw_sum, name = "w_norm")
# disp_slice = tf.slice(out_batch,[0,0],[-1,1], name = "disp_slice")
# d_gt_slice = tf.slice(gt_ds_batch,[0,0],[-1,1], name = "d_gt_slice")
disp_slice = tf.reshape(out_batch[:,0],[-1], name = "disp_slice")
d_gt_slice = tf.reshape(gt_ds_batch[:,0],[-1], name = "d_gt_slice")
if absolute_disparity:
out_diff = tf.subtract(disp_slice, d_gt_slice, name = "out_diff")
else:
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
residual_disp = tf.subtract(d_gt_slice, td_flat, name = "residual_disp")
out_diff = tf.subtract(disp_slice, residual_disp, name = "out_diff")
out_diff2 = tf.square(out_diff, name = "out_diff2")
out_wdiff2 = tf.multiply (out_diff2, w_norm, name = "out_wdiff2")
cost1 = tf.reduce_sum(out_wdiff2, name = "cost1")
out_diff2_offset = tf.subtract(out_diff2, error2_offset, name = "out_diff2_offset")
out_diff2_biased = tf.maximum(out_diff2_offset, 0.0, name = "out_diff2_biased")
out_diff2_wbiased = tf.multiply(out_diff2_biased, w_norm, name = "out_diff2_wbiased")
cost1b = tf.reduce_sum(out_diff2_wbiased, name = "cost1b")
if use_confidence:
cost12 = tf.add(cost1b, cost2, name = "cost12")
cost123 = tf.add(cost12, cost3, name = "cost123")
return cost123, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
else:
return cost1b, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
#corr2d325 = tf.concat([corr2d,target_disparity],0)
#corr2d325 = tf.concat([next_element_train['corr2d'],tf.reshape(next_element_train['target_disparity'],(-1,1))],1)
corr2d325 = tf.concat([next_element_train['corr2d'], next_element_train['target_disparity']],1)
#next_element_train
# in_features = tf.concat([corr2d,target_disparity],0)
out = network(corr2d325)
#Try standard loss functions first
G_loss, _disp_slice, _d_gt_slice, _out_diff, _out_diff2, _w_norm, _out_wdiff2, _cost1 = batchLoss(out_batch = out, # [batch_size,(1..2)] tf_result
target_disparity_batch= next_element_train['target_disparity'], # target_disparity, ### target_d, # [batch_size] tf placeholder
gt_ds_batch = next_element_train['gt_ds'], # gt_ds, ### gt, # [batch_size,2] tf placeholder
absolute_disparity = ABSOLUTE_DISPARITY,
use_confidence = USE_CONFIDENCE, # True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 2.0)
tf.summary.scalar("G_loss",G_loss)
tf.summary.scalar("sq_diff",_cost1)
t_vars=tf.trainable_variables()
lr=tf.placeholder(tf.float32)
G_opt=tf.train.AdamOptimizer(learning_rate=lr).minimize(G_loss)
saver=tf.train.Saver()
ROOT_PATH = './attic/nn_ds_inmem_graph2/'
TRAIN_PATH = ROOT_PATH + 'train'
TEST_PATH = ROOT_PATH + 'test'
# CLEAN OLD STAFF
shutil.rmtree(TRAIN_PATH, ignore_errors=True)
shutil.rmtree(TEST_PATH, ignore_errors=True)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter(TRAIN_PATH, sess.graph)
test_writer = tf.summary.FileWriter(TEST_PATH, sess.graph)
loss_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss_test_hist= np.empty(dataset_test_size, dtype=np.float32)
loss2_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss2_test_hist= np.empty(dataset_test_size, dtype=np.float32)
for epoch in range(EPOCHS_TO_RUN):
# if SHUFFLE_EPOCH:
# dataset_train = dataset_train.shuffle(buffer_size=10000)
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_train,
target_disparity_train_placeholder: target_disparity_train,
gt_ds_train_placeholder: gt_ds_train})
for i in range(dataset_train_size):
try:
train_summary,_, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[ merged,
G_opt,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR})
# save all for now as a test
#train_writer.add_summary(summary, i)
#train_writer.add_summary(train_summary, i)
loss_train_hist[i] = G_loss_trained
loss2_train_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("train done at step %d"%(i))
break
train_avg = np.average(loss_train_hist)
train2_avg = np.average(loss2_train_hist)
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_test,
target_disparity_train_placeholder: target_disparity_test,
gt_ds_train_placeholder: gt_ds_test})
for i in range(dataset_test_size):
try:
test_summary, G_loss_tested, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[merged,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR})
loss_test_hist[i] = G_loss_tested
loss2_test_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("test done at step %d"%(i))
break
# print_time("%d:%d -> %f"%(epoch,i,G_current))
test_avg = np.average(loss_test_hist)
test2_avg = np.average(loss2_test_hist)
train_writer.add_summary(train_summary, epoch)
test_writer.add_summary(test_summary, epoch)
print_time("%d:%d -> %f %f (%f %f)"%(epoch,i,train_avg, test_avg,train2_avg, test2_avg))
# Close writers
train_writer.close()
test_writer.close()
#reports error: Exception ignored in: <bound method BaseSession.__del__ of <tensorflow.python.client.session.Session object at 0x7efc5f720ef0>> if there is no print before exit()
print("All done")
exit (0)
nn_ds_inmem3.py 0 → 100644
View file @ d7fc8ac4
#!/usr/bin/env python3
from numpy import float64
__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 itertools
import time
import matplotlib.pyplot as plt
import shutil
TIME_START = time.time()
TIME_LAST = TIME_START
DEBUG_LEVEL= 1
DISP_BATCH_BINS = 20 # Number of batch disparity bins
STR_BATCH_BINS = 10 # Number of batch strength bins
FILES_PER_SCENE = 5 # number of random offset files for the scene to select from (0 - use all available)
#MIN_BATCH_CHOICES = 10 # minimal number of tiles in a file for each bin to select from
#MAX_BATCH_FILES = 10 #maximal number of files to use in a batch
MAX_EPOCH = 500
LR = 1e-4 # learning rate
USE_CONFIDENCE = False
ABSOLUTE_DISPARITY = True # False
DEBUG_PLT_LOSS = True
FEATURES_PER_TILE = 324
EPOCHS_TO_RUN = 10000 #0
RUN_TOT_AVG = 100 # last batches to average. Epoch is 307 training batches
BATCH_SIZE = 1000 # Each batch of tiles has balanced D/S tiles, shuffled batches but not inside batches
SHUFFLE_EPOCH = True
#DEBUG_PACK_TILES = True
#http://stackoverflow.com/questions/287871/print-in-terminal-with-colors-using-python
class bcolors:
HEADER = '\033[95m'
OKBLUE = '\033[94m'
OKGREEN = '\033[92m'
WARNING = '\033[38;5;214m'
FAIL = '\033[91m'
ENDC = '\033[0m'
BOLD = '\033[1m'
BOLDWHITE = '\033[1;37m'
UNDERLINE = '\033[4m'
def print_time(txt="",end="\n"):
global TIME_LAST
t = time.time()
if txt:
txt +=" "
print(("%s"+bcolors.BOLDWHITE+"at %.4fs (+%.4fs)"+bcolors.ENDC)%(txt,t-TIME_START,t-TIME_LAST), end = end, flush=True)
TIME_LAST = t
#reading to memory (testing)
def readTFRewcordsEpoch(train_filename):
# filenames = [train_filename]
# dataset = tf.data.TFRecordDataset(filenames)
if not '.tfrecords' in train_filename:
train_filename += '.tfrecords'
record_iterator = tf.python_io.tf_record_iterator(path=train_filename)
corr2d_list=[]
target_disparity_list=[]
gt_ds_list = []
for string_record in record_iterator:
example = tf.train.Example()
example.ParseFromString(string_record)
corr2d_list.append (np.array(example.features.feature['corr2d'].float_list.value, dtype=np.float32))
# target_disparity_list.append(np.array(example.features.feature['target_disparity'].float_list.value[0], dtype=np.float32))
target_disparity_list.append (np.array(example.features.feature['target_disparity'].float_list.value, dtype=np.float32))
gt_ds_list.append (np.array(example.features.feature['gt_ds'].float_list.value, dtype= np.float32))
corr2d= np.array(corr2d_list)
target_disparity = np.array(target_disparity_list)
gt_ds = np.array(gt_ds_list)
return corr2d, target_disparity, gt_ds
#from http://warmspringwinds.github.io/tensorflow/tf-slim/2016/12/21/tfrecords-guide/
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'corr2d': tf.FixedLenFeature([324],tf.float32), #string),
'target_disparity': tf.FixedLenFeature([1], tf.float32), #.string),
'gt_ds': tf.FixedLenFeature([2], tf.float32) #.string)
})
corr2d = features['corr2d'] # tf.decode_raw(features['corr2d'], tf.float32)
target_disparity = features['target_disparity'] # tf.decode_raw(features['target_disparity'], tf.float32)
gt_ds = tf.cast(features['gt_ds'], tf.float32) # tf.decode_raw(features['gt_ds'], tf.float32)
in_features = tf.concat([corr2d,target_disparity],0)
# still some nan-s in correlation data?
# in_features_clean = tf.where(tf.is_nan(in_features), tf.zeros_like(in_features), in_features)
# corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features_clean, target_disparity, gt_ds],
corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features, target_disparity, gt_ds],
batch_size=1000, # 2,
capacity=30,
num_threads=2,
min_after_dequeue=10)
return corr2d_out, target_disparity_out, gt_ds_out
#http://adventuresinmachinelearning.com/introduction-tensorflow-queuing/
#Main code
try:
train_filenameTFR = sys.argv[1]
except IndexError:
train_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/train.tfrecords"
try:
test_filenameTFR = sys.argv[2]
except IndexError:
test_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/test.tfrecords"
#FILES_PER_SCENE
import tensorflow as tf
import tensorflow.contrib.slim as slim
print_time("Importing training data... ", end="")
corr2d_train, target_disparity_train, gt_ds_train = readTFRewcordsEpoch(train_filenameTFR)
print_time(" Done")
corr2d_train_placeholder = tf.placeholder(corr2d_train.dtype, (None,324)) # corr2d_train.shape)
target_disparity_train_placeholder = tf.placeholder(target_disparity_train.dtype, (None,1)) #target_disparity_train.shape)
gt_ds_train_placeholder = tf.placeholder(gt_ds_train.dtype, (None,2)) #gt_ds_train.shape)
dataset_train = tf.data.Dataset.from_tensor_slices({
"corr2d":corr2d_train_placeholder,
"target_disparity": target_disparity_train_placeholder,
"gt_ds": gt_ds_train_placeholder})
dataset_train_size = len(corr2d_train)
print_time("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
dataset_train = dataset_train.batch(BATCH_SIZE)
dataset_train_size //= BATCH_SIZE
print("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
iterator_train = dataset_train.make_initializable_iterator()
next_element_train = iterator_train.get_next()
print_time("Importing test data... ", end="")
corr2d_test, target_disparity_test, gt_ds_test = readTFRewcordsEpoch(test_filenameTFR)
print_time(" Done")
dataset_test_size = len(corr2d_test)
dataset_test_size //= BATCH_SIZE
"""
iterator_test = dataset_test.make_initializable_iterator()
next_element_test = iterator_test.get_next()
"""
#https://www.tensorflow.org/versions/r1.5/programmers_guide/datasets
result_dir = './attic/result_inmem3/'
checkpoint_dir = './attic/result_inmem3/'
save_freq = 500
def lrelu(x):
return tf.maximum(x*0.2,x)
# return tf.nn.relu(x)
def network(input):
# fc1 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc1')
# fc2 = slim.fully_connected(fc1, 128, activation_fn=lrelu,scope='g_fc2')
## fc3 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc3')
## fc4 = slim.fully_connected(fc3, 128, activation_fn=lrelu,scope='g_fc4')
## fc5 = slim.fully_connected(fc4, 64, activation_fn=lrelu,scope='g_fc5')
fc3 = slim.fully_connected(input, 32, activation_fn=lrelu,scope='g_fc3')
fc4 = slim.fully_connected(fc3, 20, activation_fn=lrelu,scope='g_fc4')
fc5 = slim.fully_connected(fc4, 16, activation_fn=lrelu,scope='g_fc5')
if USE_CONFIDENCE:
fc6 = slim.fully_connected(fc5, 2, activation_fn=lrelu,scope='g_fc6')
else:
fc6 = slim.fully_connected(fc5, 1, activation_fn=None,scope='g_fc6')
#If using residual disparity, split last layer into 2 or remove activation and add rectifier to confidence only
return fc6
def batchLoss(out_batch, # [batch_size,(1..2)] tf_result
target_disparity_batch, # [batch_size] tf placeholder
gt_ds_batch, # [batch_size,2] tf placeholder
absolute_disparity = True, #when false there should be no activation on disparity output !
use_confidence = True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 1.0,
error2_offset = 0.0025, # (0.05^2)
disp_wmin = 1.0, # minimal disparity to apply weight boosting for small disparities
disp_wmax = 8.0, # maximal disparity to apply weight boosting for small disparities
use_out = False): # use calculated disparity for disparity weight boosting (False - use target disparity)
with tf.name_scope("BatchLoss"):
"""
Here confidence should be after relU. Disparity - may be also if absolute, but no activation if output is residual disparity
"""
tf_lambda_conf_avg = tf.constant(lambda_conf_avg, dtype=tf.float32, name="tf_lambda_conf_avg")
tf_lambda_conf_pwr = tf.constant(lambda_conf_pwr, dtype=tf.float32, name="tf_lambda_conf_pwr")
tf_conf_pwr = tf.constant(conf_pwr, dtype=tf.float32, name="tf_conf_pwr")
tf_gt_conf_offset = tf.constant(gt_conf_offset, dtype=tf.float32, name="tf_gt_conf_offset")
tf_gt_conf_pwr = tf.constant(gt_conf_pwr, dtype=tf.float32, name="tf_gt_conf_pwr")
tf_num_tiles = tf.shape(gt_ds_batch)[0]
tf_0f = tf.constant(0.0, dtype=tf.float32, name="tf_0f")
tf_1f = tf.constant(1.0, dtype=tf.float32, name="tf_1f")
tf_maxw = tf.constant(1.0, dtype=tf.float32, name="tf_maxw")
if gt_conf_pwr == 0:
w = tf.ones((out_batch.shape[0]), dtype=tf.float32,name="w_ones")
else:
# w_slice = tf.slice(gt_ds_batch,[0,1],[-1,1], name = "w_gt_slice")
w_slice = tf.reshape(gt_ds_batch[:,1],[-1], name = "w_gt_slice")
w_sub = tf.subtract (w_slice, tf_gt_conf_offset, name = "w_sub")
# w_clip = tf.clip_by_value(w_sub, tf_0f,tf_maxw, name = "w_clip")
w_clip = tf.maximum(w_sub, tf_0f, name = "w_clip")
if gt_conf_pwr == 1.0:
w = w_clip
else:
w=tf.pow(w_clip, tf_gt_conf_pwr, name = "w_pow")
if use_confidence:
tf_num_tilesf = tf.cast(tf_num_tiles, dtype=tf.float32, name="tf_num_tilesf")
# conf_slice = tf.slice(out_batch,[0,1],[-1,1], name = "conf_slice")
conf_slice = tf.reshape(out_batch[:,1],[-1], name = "conf_slice")
conf_sum = tf.reduce_sum(conf_slice, name = "conf_sum")
conf_avg = tf.divide(conf_sum, tf_num_tilesf, name = "conf_avg")
conf_avg1 = tf.subtract(conf_avg, tf_1f, name = "conf_avg1")
conf_avg2 = tf.square(conf_avg1, name = "conf_avg2")
cost2 = tf.multiply (conf_avg2, tf_lambda_conf_avg, name = "cost2")
iconf_avg = tf.divide(tf_1f, conf_avg, name = "iconf_avg")
nconf = tf.multiply (conf_slice, iconf_avg, name = "nconf") #normalized confidence
nconf_pwr = tf.pow(nconf, conf_pwr, name = "nconf_pwr")
nconf_pwr_sum = tf.reduce_sum(nconf_pwr, name = "nconf_pwr_sum")
nconf_pwr_offs = tf.subtract(nconf_pwr_sum, tf_1f, name = "nconf_pwr_offs")
cost3 = tf.multiply (conf_avg2, nconf_pwr_offs, name = "cost3")
w_all = tf.multiply (w, nconf, name = "w_all")
else:
w_all = w
# cost2 = 0.0
# cost3 = 0.0
# normalize weights
w_sum = tf.reduce_sum(w_all, name = "w_sum")
iw_sum = tf.divide(tf_1f, w_sum, name = "iw_sum")
w_norm = tf.multiply (w_all, iw_sum, name = "w_norm")
# disp_slice = tf.slice(out_batch,[0,0],[-1,1], name = "disp_slice")
# d_gt_slice = tf.slice(gt_ds_batch,[0,0],[-1,1], name = "d_gt_slice")
disp_slice = tf.reshape(out_batch[:,0],[-1], name = "disp_slice")
d_gt_slice = tf.reshape(gt_ds_batch[:,0],[-1], name = "d_gt_slice")
"""
if absolute_disparity:
out_diff = tf.subtract(disp_slice, d_gt_slice, name = "out_diff")
else:
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
residual_disp = tf.subtract(d_gt_slice, td_flat, name = "residual_disp")
out_diff = tf.subtract(disp_slice, residual_disp, name = "out_diff")
"""
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
if absolute_disparity:
adisp = disp_slice
else:
# td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
adisp = tf.add(disp_slice, td_flat, name - "adisp")
out_diff = tf.subtract(adisp, d_gt_slice, name = "out_diff")
out_diff2 = tf.square(out_diff, name = "out_diff2")
out_wdiff2 = tf.multiply (out_diff2, w_norm, name = "out_wdiff2")
cost1 = tf.reduce_sum(out_wdiff2, name = "cost1")
out_diff2_offset = tf.subtract(out_diff2, error2_offset, name = "out_diff2_offset")
out_diff2_biased = tf.maximum(out_diff2_offset, 0.0, name = "out_diff2_biased")
# calculate disparity-based weight boost
if use_out:
dispw = tf.clip_by_value(adisp, disp_wmin, disp_wmax, name = "dispw")
else:
dispw = tf.clip_by_value(td_flat, disp_wmin, disp_wmax, name = "dispw")
dispw_boost = tf.divide(disp_wmax, dispw, name = "dispw_boost")
dispw_comp = tf.multiply (dispw_boost, w_norm, name = "dispw_comp")
dispw_sum = tf.reduce_sum(dispw_comp, name = "dispw_sum")
idispw_sum = tf.divide(tf_1f, dispw_sum, name = "idispw_sum")
dispw_norm = tf.multiply (dispw_comp, idispw_sum, name = "dispw_norm")
out_diff2_wbiased = tf.multiply(out_diff2_biased, dispw_norm, name = "out_diff2_wbiased")
# out_diff2_wbiased = tf.multiply(out_diff2_biased, w_norm, name = "out_diff2_wbiased")
cost1b = tf.reduce_sum(out_diff2_wbiased, name = "cost1b")
if use_confidence:
cost12 = tf.add(cost1b, cost2, name = "cost12")
cost123 = tf.add(cost12, cost3, name = "cost123")
return cost123, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
else:
return cost1b, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
#corr2d325 = tf.concat([corr2d,target_disparity],0)
#corr2d325 = tf.concat([next_element_train['corr2d'],tf.reshape(next_element_train['target_disparity'],(-1,1))],1)
corr2d325 = tf.concat([next_element_train['corr2d'], next_element_train['target_disparity']],1)
#next_element_train
# in_features = tf.concat([corr2d,target_disparity],0)
out = network(corr2d325)
#Try standard loss functions first
G_loss, _disp_slice, _d_gt_slice, _out_diff, _out_diff2, _w_norm, _out_wdiff2, _cost1 = batchLoss(out_batch = out, # [batch_size,(1..2)] tf_result
target_disparity_batch= next_element_train['target_disparity'], # target_disparity, ### target_d, # [batch_size] tf placeholder
gt_ds_batch = next_element_train['gt_ds'], # gt_ds, ### gt, # [batch_size,2] tf placeholder
absolute_disparity = ABSOLUTE_DISPARITY,
use_confidence = USE_CONFIDENCE, # True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 2.0,
error2_offset = 0.0025, # (0.05^2)
disp_wmin = 1.0, # minimal disparity to apply weight boosting for small disparities
disp_wmax = 8.0, # maximal disparity to apply weight boosting for small disparities
use_out = False) # use calculated disparity for disparity weight boosting (False - use target disparity)
tf_ph_G_loss = tf.placeholder(tf.float32,shape=None,name='G_loss_avg')
tf_ph_sq_diff = tf.placeholder(tf.float32,shape=None,name='sq_diff_avg')
with tf.name_scope('sample'):
tf.summary.scalar("G_loss",G_loss)
tf.summary.scalar("sq_diff",_cost1)
with tf.name_scope('epoch_average'):
tf.summary.scalar("G_loss_epoch",tf_ph_G_loss)
tf.summary.scalar("sq_diff_epoch",tf_ph_sq_diff)
t_vars=tf.trainable_variables()
lr=tf.placeholder(tf.float32)
G_opt=tf.train.AdamOptimizer(learning_rate=lr).minimize(G_loss)
saver=tf.train.Saver()
ROOT_PATH = './attic/nn_ds_inmem_graph3/'
TRAIN_PATH = ROOT_PATH + 'train'
TEST_PATH = ROOT_PATH + 'test'
# CLEAN OLD STAFF
shutil.rmtree(TRAIN_PATH, ignore_errors=True)
shutil.rmtree(TEST_PATH, ignore_errors=True)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter(TRAIN_PATH, sess.graph)
test_writer = tf.summary.FileWriter(TEST_PATH, sess.graph)
loss_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss_test_hist= np.empty(dataset_test_size, dtype=np.float32)
loss2_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss2_test_hist= np.empty(dataset_test_size, dtype=np.float32)
train_avg = 0.0
train2_avg = 0.0
test_avg = 0.0
test2_avg = 0.0
for epoch in range(EPOCHS_TO_RUN):
# if SHUFFLE_EPOCH:
# dataset_train = dataset_train.shuffle(buffer_size=10000)
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_train,
target_disparity_train_placeholder: target_disparity_train,
gt_ds_train_placeholder: gt_ds_train})
for i in range(dataset_train_size):
try:
train_summary,_, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[ merged,
G_opt,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR,tf_ph_G_loss:train_avg, tf_ph_sq_diff:train2_avg}) # pfrevious value of *_avg
# save all for now as a test
#train_writer.add_summary(summary, i)
#train_writer.add_summary(train_summary, i)
loss_train_hist[i] = G_loss_trained
loss2_train_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("train done at step %d"%(i))
break
train_avg = np.average(loss_train_hist).astype(np.float32)
train2_avg = np.average(loss2_train_hist).astype(np.float32)
# _,_=sess.run([tf_ph_G_loss,tf_ph_sq_diff],feed_dict={tf_ph_G_loss:train_avg, tf_ph_sq_diff:train2_avg})
#tf_ph_G_loss = tf.placeholder(tf.float32,shape=None,name='G_loss_avg')
#tf_ph_sq_diff = tf.placeholder(tf.float32,shape=None,name='sq_diff_avg')
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_test,
target_disparity_train_placeholder: target_disparity_test,
gt_ds_train_placeholder: gt_ds_test})
for i in range(dataset_test_size):
try:
test_summary, G_loss_tested, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[merged,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR,tf_ph_G_loss:test_avg, tf_ph_sq_diff:test2_avg}) # pfrevious value of *_avg
loss_test_hist[i] = G_loss_tested
loss2_test_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("test done at step %d"%(i))
break
# print_time("%d:%d -> %f"%(epoch,i,G_current))
test_avg = np.average(loss_test_hist).astype(np.float32)
test2_avg = np.average(loss2_test_hist).astype(np.float32)
# _,_=sess.run([tf_ph_G_loss,tf_ph_sq_diff],feed_dict={tf_ph_G_loss:test_avg, tf_ph_sq_diff:test2_avg})
train_writer.add_summary(train_summary, epoch)
test_writer.add_summary(test_summary, epoch)
print_time("%d:%d -> %f %f (%f %f)"%(epoch,i,train_avg, test_avg,train2_avg, test2_avg))
# Close writers
train_writer.close()
test_writer.close()
#reports error: Exception ignored in: <bound method BaseSession.__del__ of <tensorflow.python.client.session.Session object at 0x7efc5f720ef0>> if there is no print before exit()
print("All done")
exit (0)
nn_ds_inmem4.py 0 → 100644
View file @ d7fc8ac4
#!/usr/bin/env python3
from numpy import float64
__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 itertools
import time
import matplotlib.pyplot as plt
import shutil
TIME_START = time.time()
TIME_LAST = TIME_START
DEBUG_LEVEL= 1
DISP_BATCH_BINS = 20 # Number of batch disparity bins
STR_BATCH_BINS = 10 # Number of batch strength bins
FILES_PER_SCENE = 5 # number of random offset files for the scene to select from (0 - use all available)
#MIN_BATCH_CHOICES = 10 # minimal number of tiles in a file for each bin to select from
#MAX_BATCH_FILES = 10 #maximal number of files to use in a batch
MAX_EPOCH = 500
LR = 1e-4 # learning rate
USE_CONFIDENCE = False
ABSOLUTE_DISPARITY = False # True # False
DEBUG_PLT_LOSS = True
FEATURES_PER_TILE = 324
EPOCHS_TO_RUN = 10000 #0
RUN_TOT_AVG = 100 # last batches to average. Epoch is 307 training batches
BATCH_SIZE = 1000 # Each batch of tiles has balanced D/S tiles, shuffled batches but not inside batches
SHUFFLE_EPOCH = True
NET_ARCH = 3 # overwrite with argv?
#DEBUG_PACK_TILES = True
SUFFIX=str(NET_ARCH)+ (["R","A"][ABSOLUTE_DISPARITY])
#http://stackoverflow.com/questions/287871/print-in-terminal-with-colors-using-python
class bcolors:
HEADER = '\033[95m'
OKBLUE = '\033[94m'
OKGREEN = '\033[92m'
WARNING = '\033[38;5;214m'
FAIL = '\033[91m'
ENDC = '\033[0m'
BOLD = '\033[1m'
BOLDWHITE = '\033[1;37m'
UNDERLINE = '\033[4m'
def print_time(txt="",end="\n"):
global TIME_LAST
t = time.time()
if txt:
txt +=" "
print(("%s"+bcolors.BOLDWHITE+"at %.4fs (+%.4fs)"+bcolors.ENDC)%(txt,t-TIME_START,t-TIME_LAST), end = end, flush=True)
TIME_LAST = t
#reading to memory (testing)
def readTFRewcordsEpoch(train_filename):
# filenames = [train_filename]
# dataset = tf.data.TFRecordDataset(filenames)
if not '.tfrecords' in train_filename:
train_filename += '.tfrecords'
record_iterator = tf.python_io.tf_record_iterator(path=train_filename)
corr2d_list=[]
target_disparity_list=[]
gt_ds_list = []
for string_record in record_iterator:
example = tf.train.Example()
example.ParseFromString(string_record)
corr2d_list.append (np.array(example.features.feature['corr2d'].float_list.value, dtype=np.float32))
# target_disparity_list.append(np.array(example.features.feature['target_disparity'].float_list.value[0], dtype=np.float32))
target_disparity_list.append (np.array(example.features.feature['target_disparity'].float_list.value, dtype=np.float32))
gt_ds_list.append (np.array(example.features.feature['gt_ds'].float_list.value, dtype= np.float32))
corr2d= np.array(corr2d_list)
target_disparity = np.array(target_disparity_list)
gt_ds = np.array(gt_ds_list)
return corr2d, target_disparity, gt_ds
#from http://warmspringwinds.github.io/tensorflow/tf-slim/2016/12/21/tfrecords-guide/
def read_and_decode(filename_queue):
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'corr2d': tf.FixedLenFeature([324],tf.float32), #string),
'target_disparity': tf.FixedLenFeature([1], tf.float32), #.string),
'gt_ds': tf.FixedLenFeature([2], tf.float32) #.string)
})
corr2d = features['corr2d'] # tf.decode_raw(features['corr2d'], tf.float32)
target_disparity = features['target_disparity'] # tf.decode_raw(features['target_disparity'], tf.float32)
gt_ds = tf.cast(features['gt_ds'], tf.float32) # tf.decode_raw(features['gt_ds'], tf.float32)
in_features = tf.concat([corr2d,target_disparity],0)
# still some nan-s in correlation data?
# in_features_clean = tf.where(tf.is_nan(in_features), tf.zeros_like(in_features), in_features)
# corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features_clean, target_disparity, gt_ds],
corr2d_out, target_disparity_out, gt_ds_out = tf.train.shuffle_batch( [in_features, target_disparity, gt_ds],
batch_size=1000, # 2,
capacity=30,
num_threads=2,
min_after_dequeue=10)
return corr2d_out, target_disparity_out, gt_ds_out
#http://adventuresinmachinelearning.com/introduction-tensorflow-queuing/
#Main code
try:
train_filenameTFR = sys.argv[1]
except IndexError:
train_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/train.tfrecords"
try:
test_filenameTFR = sys.argv[2]
except IndexError:
test_filenameTFR = "/mnt/dde6f983-d149-435e-b4a2-88749245cc6c/home/eyesis/x3d_data/data_sets/tf_data/test.tfrecords"
#FILES_PER_SCENE
import tensorflow as tf
import tensorflow.contrib.slim as slim
print_time("Importing training data... ", end="")
corr2d_train, target_disparity_train, gt_ds_train = readTFRewcordsEpoch(train_filenameTFR)
print_time(" Done")
corr2d_train_placeholder = tf.placeholder(corr2d_train.dtype, (None,324)) # corr2d_train.shape)
target_disparity_train_placeholder = tf.placeholder(target_disparity_train.dtype, (None,1)) #target_disparity_train.shape)
gt_ds_train_placeholder = tf.placeholder(gt_ds_train.dtype, (None,2)) #gt_ds_train.shape)
dataset_train = tf.data.Dataset.from_tensor_slices({
"corr2d":corr2d_train_placeholder,
"target_disparity": target_disparity_train_placeholder,
"gt_ds": gt_ds_train_placeholder})
dataset_train_size = len(corr2d_train)
print_time("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
dataset_train = dataset_train.batch(BATCH_SIZE)
dataset_train_size //= BATCH_SIZE
print("dataset_train.output_types "+str(dataset_train.output_types)+", dataset_train.output_shapes "+str(dataset_train.output_shapes)+", number of elements="+str(dataset_train_size))
iterator_train = dataset_train.make_initializable_iterator()
next_element_train = iterator_train.get_next()
print_time("Importing test data... ", end="")
corr2d_test, target_disparity_test, gt_ds_test = readTFRewcordsEpoch(test_filenameTFR)
print_time(" Done")
dataset_test_size = len(corr2d_test)
dataset_test_size //= BATCH_SIZE
"""
iterator_test = dataset_test.make_initializable_iterator()
next_element_test = iterator_test.get_next()
"""
#https://www.tensorflow.org/versions/r1.5/programmers_guide/datasets
result_dir = './attic/result_inmem4_'+ SUFFIX+'/'
checkpoint_dir = './attic/result_inmem4_'+ SUFFIX+'/'
save_freq = 500
def lrelu(x):
return tf.maximum(x*0.2,x)
# return tf.nn.relu(x)
def network_fc_simple(input, arch = 0):
layouts = {0:[0, 0, 0, 32, 20, 16],
1:[0, 0, 0, 256, 128, 64],
2:[0, 128, 32, 32, 32, 16],
3:[0, 0, 40, 32, 20, 16]}
layout = layouts[arch]
last_indx = None;
fc = []
for i, num_outs in enumerate (layout):
if num_outs:
if fc:
inp = fc[-1]
else:
inp = input
fc.append(slim.fully_connected(inp, num_outs, activation_fn=lrelu,scope='g_fc'+str(i)))
"""
# fc1 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc1')
# fc2 = slim.fully_connected(fc1, 128, activation_fn=lrelu,scope='g_fc2')
fc3 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc3')
fc4 = slim.fully_connected(fc3, 128, activation_fn=lrelu,scope='g_fc4')
fc5 = slim.fully_connected(fc4, 64, activation_fn=lrelu,scope='g_fc5')
"""
### fc3 = slim.fully_connected(input, 32, activation_fn=lrelu,scope='g_fc3')
### fc4 = slim.fully_connected(fc3, 20, activation_fn=lrelu,scope='g_fc4')
### fc5 = slim.fully_connected(fc4, 16, activation_fn=lrelu,scope='g_fc5')
if USE_CONFIDENCE:
fc_out = slim.fully_connected(fc[-1], 2, activation_fn=lrelu,scope='g_fc_out')
else:
fc_out = slim.fully_connected(fc[-1], 1, activation_fn=None,scope='g_fc_out')
#If using residual disparity, split last layer into 2 or remove activation and add rectifier to confidence only
return fc_out
def batchLoss(out_batch, # [batch_size,(1..2)] tf_result
target_disparity_batch, # [batch_size] tf placeholder
gt_ds_batch, # [batch_size,2] tf placeholder
absolute_disparity = True, #when false there should be no activation on disparity output !
use_confidence = True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 1.0,
error2_offset = 0.0025, # (0.05^2)
disp_wmin = 1.0, # minimal disparity to apply weight boosting for small disparities
disp_wmax = 8.0, # maximal disparity to apply weight boosting for small disparities
use_out = False): # use calculated disparity for disparity weight boosting (False - use target disparity)
with tf.name_scope("BatchLoss"):
"""
Here confidence should be after relU. Disparity - may be also if absolute, but no activation if output is residual disparity
"""
tf_lambda_conf_avg = tf.constant(lambda_conf_avg, dtype=tf.float32, name="tf_lambda_conf_avg")
tf_lambda_conf_pwr = tf.constant(lambda_conf_pwr, dtype=tf.float32, name="tf_lambda_conf_pwr")
tf_conf_pwr = tf.constant(conf_pwr, dtype=tf.float32, name="tf_conf_pwr")
tf_gt_conf_offset = tf.constant(gt_conf_offset, dtype=tf.float32, name="tf_gt_conf_offset")
tf_gt_conf_pwr = tf.constant(gt_conf_pwr, dtype=tf.float32, name="tf_gt_conf_pwr")
tf_num_tiles = tf.shape(gt_ds_batch)[0]
tf_0f = tf.constant(0.0, dtype=tf.float32, name="tf_0f")
tf_1f = tf.constant(1.0, dtype=tf.float32, name="tf_1f")
tf_maxw = tf.constant(1.0, dtype=tf.float32, name="tf_maxw")
if gt_conf_pwr == 0:
w = tf.ones((out_batch.shape[0]), dtype=tf.float32,name="w_ones")
else:
# w_slice = tf.slice(gt_ds_batch,[0,1],[-1,1], name = "w_gt_slice")
w_slice = tf.reshape(gt_ds_batch[:,1],[-1], name = "w_gt_slice")
w_sub = tf.subtract (w_slice, tf_gt_conf_offset, name = "w_sub")
# w_clip = tf.clip_by_value(w_sub, tf_0f,tf_maxw, name = "w_clip")
w_clip = tf.maximum(w_sub, tf_0f, name = "w_clip")
if gt_conf_pwr == 1.0:
w = w_clip
else:
w=tf.pow(w_clip, tf_gt_conf_pwr, name = "w_pow")
if use_confidence:
tf_num_tilesf = tf.cast(tf_num_tiles, dtype=tf.float32, name="tf_num_tilesf")
# conf_slice = tf.slice(out_batch,[0,1],[-1,1], name = "conf_slice")
conf_slice = tf.reshape(out_batch[:,1],[-1], name = "conf_slice")
conf_sum = tf.reduce_sum(conf_slice, name = "conf_sum")
conf_avg = tf.divide(conf_sum, tf_num_tilesf, name = "conf_avg")
conf_avg1 = tf.subtract(conf_avg, tf_1f, name = "conf_avg1")
conf_avg2 = tf.square(conf_avg1, name = "conf_avg2")
cost2 = tf.multiply (conf_avg2, tf_lambda_conf_avg, name = "cost2")
iconf_avg = tf.divide(tf_1f, conf_avg, name = "iconf_avg")
nconf = tf.multiply (conf_slice, iconf_avg, name = "nconf") #normalized confidence
nconf_pwr = tf.pow(nconf, conf_pwr, name = "nconf_pwr")
nconf_pwr_sum = tf.reduce_sum(nconf_pwr, name = "nconf_pwr_sum")
nconf_pwr_offs = tf.subtract(nconf_pwr_sum, tf_1f, name = "nconf_pwr_offs")
cost3 = tf.multiply (conf_avg2, nconf_pwr_offs, name = "cost3")
w_all = tf.multiply (w, nconf, name = "w_all")
else:
w_all = w
# cost2 = 0.0
# cost3 = 0.0
# normalize weights
w_sum = tf.reduce_sum(w_all, name = "w_sum")
iw_sum = tf.divide(tf_1f, w_sum, name = "iw_sum")
w_norm = tf.multiply (w_all, iw_sum, name = "w_norm")
# disp_slice = tf.slice(out_batch,[0,0],[-1,1], name = "disp_slice")
# d_gt_slice = tf.slice(gt_ds_batch,[0,0],[-1,1], name = "d_gt_slice")
disp_slice = tf.reshape(out_batch[:,0],[-1], name = "disp_slice")
d_gt_slice = tf.reshape(gt_ds_batch[:,0],[-1], name = "d_gt_slice")
"""
if absolute_disparity:
out_diff = tf.subtract(disp_slice, d_gt_slice, name = "out_diff")
else:
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
residual_disp = tf.subtract(d_gt_slice, td_flat, name = "residual_disp")
out_diff = tf.subtract(disp_slice, residual_disp, name = "out_diff")
"""
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
if absolute_disparity:
adisp = disp_slice
else:
# td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat")
adisp = tf.add(disp_slice, td_flat, name = "adisp")
out_diff = tf.subtract(adisp, d_gt_slice, name = "out_diff")
out_diff2 = tf.square(out_diff, name = "out_diff2")
out_wdiff2 = tf.multiply (out_diff2, w_norm, name = "out_wdiff2")
cost1 = tf.reduce_sum(out_wdiff2, name = "cost1")
out_diff2_offset = tf.subtract(out_diff2, error2_offset, name = "out_diff2_offset")
out_diff2_biased = tf.maximum(out_diff2_offset, 0.0, name = "out_diff2_biased")
# calculate disparity-based weight boost
if use_out:
dispw = tf.clip_by_value(adisp, disp_wmin, disp_wmax, name = "dispw")
else:
dispw = tf.clip_by_value(td_flat, disp_wmin, disp_wmax, name = "dispw")
dispw_boost = tf.divide(disp_wmax, dispw, name = "dispw_boost")
dispw_comp = tf.multiply (dispw_boost, w_norm, name = "dispw_comp")
dispw_sum = tf.reduce_sum(dispw_comp, name = "dispw_sum")
idispw_sum = tf.divide(tf_1f, dispw_sum, name = "idispw_sum")
dispw_norm = tf.multiply (dispw_comp, idispw_sum, name = "dispw_norm")
out_diff2_wbiased = tf.multiply(out_diff2_biased, dispw_norm, name = "out_diff2_wbiased")
# out_diff2_wbiased = tf.multiply(out_diff2_biased, w_norm, name = "out_diff2_wbiased")
cost1b = tf.reduce_sum(out_diff2_wbiased, name = "cost1b")
if use_confidence:
cost12 = tf.add(cost1b, cost2, name = "cost12")
cost123 = tf.add(cost12, cost3, name = "cost123")
return cost123, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
else:
return cost1b, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
#corr2d325 = tf.concat([corr2d,target_disparity],0)
#corr2d325 = tf.concat([next_element_train['corr2d'],tf.reshape(next_element_train['target_disparity'],(-1,1))],1)
corr2d325 = tf.concat([next_element_train['corr2d'], next_element_train['target_disparity']],1)
#next_element_train
# in_features = tf.concat([corr2d,target_disparity],0)
out = network_fc_simple(input=corr2d325, arch = NET_ARCH)
#Try standard loss functions first
G_loss, _disp_slice, _d_gt_slice, _out_diff, _out_diff2, _w_norm, _out_wdiff2, _cost1 = batchLoss(out_batch = out, # [batch_size,(1..2)] tf_result
target_disparity_batch= next_element_train['target_disparity'], # target_disparity, ### target_d, # [batch_size] tf placeholder
gt_ds_batch = next_element_train['gt_ds'], # gt_ds, ### gt, # [batch_size,2] tf placeholder
absolute_disparity = ABSOLUTE_DISPARITY,
use_confidence = USE_CONFIDENCE, # True,
lambda_conf_avg = 0.01,
lambda_conf_pwr = 0.1,
conf_pwr = 2.0,
gt_conf_offset = 0.08,
gt_conf_pwr = 2.0,
error2_offset = 0.0025, # (0.05^2)
disp_wmin = 1.0, # minimal disparity to apply weight boosting for small disparities
disp_wmax = 8.0, # maximal disparity to apply weight boosting for small disparities
use_out = False) # use calculated disparity for disparity weight boosting (False - use target disparity)
tf_ph_G_loss = tf.placeholder(tf.float32,shape=None,name='G_loss_avg')
tf_ph_sq_diff = tf.placeholder(tf.float32,shape=None,name='sq_diff_avg')
with tf.name_scope('sample'):
tf.summary.scalar("G_loss",G_loss)
tf.summary.scalar("sq_diff",_cost1)
with tf.name_scope('epoch_average'):
tf.summary.scalar("G_loss_epoch",tf_ph_G_loss)
tf.summary.scalar("sq_diff_epoch",tf_ph_sq_diff)
t_vars=tf.trainable_variables()
lr=tf.placeholder(tf.float32)
G_opt=tf.train.AdamOptimizer(learning_rate=lr).minimize(G_loss)
saver=tf.train.Saver()
ROOT_PATH = './attic/nn_ds_inmem4_graph'+SUFFIX+"/"
TRAIN_PATH = ROOT_PATH + 'train'
TEST_PATH = ROOT_PATH + 'test'
# CLEAN OLD STAFF
shutil.rmtree(TRAIN_PATH, ignore_errors=True)
shutil.rmtree(TEST_PATH, ignore_errors=True)
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
merged = tf.summary.merge_all()
train_writer = tf.summary.FileWriter(TRAIN_PATH, sess.graph)
test_writer = tf.summary.FileWriter(TEST_PATH, sess.graph)
loss_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss_test_hist= np.empty(dataset_test_size, dtype=np.float32)
loss2_train_hist= np.empty(dataset_train_size, dtype=np.float32)
loss2_test_hist= np.empty(dataset_test_size, dtype=np.float32)
train_avg = 0.0
train2_avg = 0.0
test_avg = 0.0
test2_avg = 0.0
for epoch in range(EPOCHS_TO_RUN):
# if SHUFFLE_EPOCH:
# dataset_train = dataset_train.shuffle(buffer_size=10000)
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_train,
target_disparity_train_placeholder: target_disparity_train,
gt_ds_train_placeholder: gt_ds_train})
for i in range(dataset_train_size):
try:
train_summary,_, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[ merged,
G_opt,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR,tf_ph_G_loss:train_avg, tf_ph_sq_diff:train2_avg}) # pfrevious value of *_avg
# save all for now as a test
#train_writer.add_summary(summary, i)
#train_writer.add_summary(train_summary, i)
loss_train_hist[i] = G_loss_trained
loss2_train_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("train done at step %d"%(i))
break
train_avg = np.average(loss_train_hist).astype(np.float32)
train2_avg = np.average(loss2_train_hist).astype(np.float32)
# _,_=sess.run([tf_ph_G_loss,tf_ph_sq_diff],feed_dict={tf_ph_G_loss:train_avg, tf_ph_sq_diff:train2_avg})
#tf_ph_G_loss = tf.placeholder(tf.float32,shape=None,name='G_loss_avg')
#tf_ph_sq_diff = tf.placeholder(tf.float32,shape=None,name='sq_diff_avg')
sess.run(iterator_train.initializer, feed_dict={corr2d_train_placeholder: corr2d_test,
target_disparity_train_placeholder: target_disparity_test,
gt_ds_train_placeholder: gt_ds_test})
for i in range(dataset_test_size):
try:
test_summary, G_loss_tested, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[merged,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
],
feed_dict={lr:LR,tf_ph_G_loss:test_avg, tf_ph_sq_diff:test2_avg}) # pfrevious value of *_avg
loss_test_hist[i] = G_loss_tested
loss2_test_hist[i] = out_cost1
except tf.errors.OutOfRangeError:
print("test done at step %d"%(i))
break
# print_time("%d:%d -> %f"%(epoch,i,G_current))
test_avg = np.average(loss_test_hist).astype(np.float32)
test2_avg = np.average(loss2_test_hist).astype(np.float32)
# _,_=sess.run([tf_ph_G_loss,tf_ph_sq_diff],feed_dict={tf_ph_G_loss:test_avg, tf_ph_sq_diff:test2_avg})
train_writer.add_summary(train_summary, epoch)
test_writer.add_summary(test_summary, epoch)
print_time("%d:%d -> %f %f (%f %f)"%(epoch,i,train_avg, test_avg,train2_avg, test2_avg))
# Close writers
train_writer.close()
test_writer.close()
#reports error: Exception ignored in: <bound method BaseSession.__del__ of <tensorflow.python.client.session.Session object at 0x7efc5f720ef0>> if there is no print before exit()
print("All done")
exit (0)
nn_ds_inmem_tmp.py
View file @ d7fc8ac4
...@@ -30,7 +30,7 @@ FILES_PER_SCENE = 5 # number of random offset files for the scene to select f ...@@ -30,7 +30,7 @@ FILES_PER_SCENE = 5 # number of random offset files for the scene to select f
#MIN_BATCH_CHOICES = 10 # minimal number of tiles in a file for each bin to select from #MIN_BATCH_CHOICES = 10 # minimal number of tiles in a file for each bin to select from
#MAX_BATCH_FILES = 10 #maximal number of files to use in a batch #MAX_BATCH_FILES = 10 #maximal number of files to use in a batch
MAX_EPOCH = 500 MAX_EPOCH = 500
LR = 1e-3 # learning rate LR = 1e-4 # learning rate
USE_CONFIDENCE = False USE_CONFIDENCE = False
ABSOLUTE_DISPARITY = False ABSOLUTE_DISPARITY = False
DEBUG_PLT_LOSS = True DEBUG_PLT_LOSS = True
...@@ -167,16 +167,21 @@ checkpoint_dir = './attic/result_inmem/' ...@@ -167,16 +167,21 @@ checkpoint_dir = './attic/result_inmem/'
save_freq = 500 save_freq = 500
def lrelu(x): def lrelu(x):
return tf.maximum(x*0.2,x) return tf.maximum(x*0.5,x)
# return tf.nn.relu(x) # return tf.nn.relu(x)
def network(input): def network(input):
# fc1 = slim.fully_connected(input, 512, activation_fn=lrelu,scope='g_fc1') fc1 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc1')
# fc2 = slim.fully_connected(fc1, 512, activation_fn=lrelu,scope='g_fc2') fc2 = slim.fully_connected(fc1, 128, activation_fn=lrelu,scope='g_fc2')
fc3 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc3') ## fc3 = slim.fully_connected(input, 256, activation_fn=lrelu,scope='g_fc3')
fc4 = slim.fully_connected(fc3, 128, activation_fn=lrelu,scope='g_fc4') ## fc4 = slim.fully_connected(fc3, 128, activation_fn=lrelu,scope='g_fc4')
fc5 = slim.fully_connected(fc4, 64, activation_fn=lrelu,scope='g_fc5') ## fc5 = slim.fully_connected(fc4, 64, activation_fn=lrelu,scope='g_fc5')
fc3 = slim.fully_connected(fc2, 64, activation_fn=lrelu,scope='g_fc3')
fc4 = slim.fully_connected(fc3, 20, activation_fn=lrelu,scope='g_fc4')
fc5 = slim.fully_connected(fc4, 16, activation_fn=lrelu,scope='g_fc5')
if USE_CONFIDENCE: if USE_CONFIDENCE:
fc6 = slim.fully_connected(fc5, 2, activation_fn=lrelu,scope='g_fc6') fc6 = slim.fully_connected(fc5, 2, activation_fn=lrelu,scope='g_fc6')
else: else:
...@@ -318,7 +323,7 @@ with tf.Session() as sess: ...@@ -318,7 +323,7 @@ with tf.Session() as sess:
for epoch in range(EPOCHS_TO_RUN): for epoch in range(EPOCHS_TO_RUN):
if SHUFFLE_EPOCH: # if SHUFFLE_EPOCH:
dataset_train = dataset_train.shuffle(buffer_size=10000) dataset_train = dataset_train.shuffle(buffer_size=10000)
sess.run(iterator_train.initializer) sess.run(iterator_train.initializer)
...@@ -331,6 +336,8 @@ with tf.Session() as sess: ...@@ -331,6 +336,8 @@ with tf.Session() as sess:
# Train run # Train run
if i<START_TEST: if i<START_TEST:
if (epoch <50) or (epoch > 100) :
try: try:
# _, G_current, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out, target_disparity_out, gt_ds_out = sess.run( # _, G_current, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out, target_disparity_out, gt_ds_out = sess.run(
train_summary,_, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run( train_summary,_, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
...@@ -347,6 +354,33 @@ with tf.Session() as sess: ...@@ -347,6 +354,33 @@ with tf.Session() as sess:
_cost1, _cost1,
corr2d325, corr2d325,
# target_disparity, # target_disparity,
# gt_ds
],
feed_dict={lr:LR})
# save all for now as a test
#train_writer.add_summary(summary, i)
#train_writer.add_summary(train_summary, i)
except tf.errors.OutOfRangeError:
break
else:
try:
# _, G_current, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out, target_disparity_out, gt_ds_out = sess.run(
train_summary, G_loss_trained, output, disp_slice, d_gt_slice, out_diff, out_diff2, w_norm, out_wdiff2, out_cost1, corr2d325_out = sess.run(
[ merged,
# G_opt,
G_loss,
out,
_disp_slice,
_d_gt_slice,
_out_diff,
_out_diff2,
_w_norm,
_out_wdiff2,
_cost1,
corr2d325,
# target_disparity,
# gt_ds # gt_ds
], ],
feed_dict={lr:LR}) feed_dict={lr:LR})
......
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