Skip to content

P
python3-imagej-tiff
Commit d0bbd388 authored by Andrey Filippov's avatar Andrey Filippov
Browse files
Options

Added scope

parent da662515
Hide whitespace changes
Inline Side-by-side
Showing with 70 additions and 69 deletions
nn_ds_inmem.py
View file @ d0bbd388
...@@ -194,77 +194,78 @@ def batchLoss(out_batch, # [batch_size,(1..2)] tf_result ...@@ -194,77 +194,78 @@ def batchLoss(out_batch, # [batch_size,(1..2)] tf_result
conf_pwr = 2.0, conf_pwr = 2.0,
gt_conf_offset = 0.08, gt_conf_offset = 0.08,
gt_conf_pwr = 1.0): gt_conf_pwr = 1.0):
""" 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 """
""" 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_lambda_conf_avg = tf.constant(lambda_conf_avg, dtype=tf.float32, name="tf_lambda_conf_avg")
tf_conf_pwr = tf.constant(conf_pwr, dtype=tf.float32, name="tf_conf_pwr") tf_lambda_conf_pwr = tf.constant(lambda_conf_pwr, dtype=tf.float32, name="tf_lambda_conf_pwr")
tf_gt_conf_offset = tf.constant(gt_conf_offset, dtype=tf.float32, name="tf_gt_conf_offset") tf_conf_pwr = tf.constant(conf_pwr, dtype=tf.float32, name="tf_conf_pwr")
tf_gt_conf_pwr = tf.constant(gt_conf_pwr, dtype=tf.float32, name="tf_gt_conf_pwr") tf_gt_conf_offset = tf.constant(gt_conf_offset, dtype=tf.float32, name="tf_gt_conf_offset")
tf_num_tiles = tf.shape(gt_ds_batch)[0] tf_gt_conf_pwr = tf.constant(gt_conf_pwr, dtype=tf.float32, name="tf_gt_conf_pwr")
tf_0f = tf.constant(0.0, dtype=tf.float32, name="tf_0f") tf_num_tiles = tf.shape(gt_ds_batch)[0]
tf_1f = tf.constant(1.0, dtype=tf.float32, name="tf_1f") tf_0f = tf.constant(0.0, dtype=tf.float32, name="tf_0f")
tf_maxw = tf.constant(1.0, dtype=tf.float32, name="tf_maxw") tf_1f = tf.constant(1.0, dtype=tf.float32, name="tf_1f")
if gt_conf_pwr == 0: tf_maxw = tf.constant(1.0, dtype=tf.float32, name="tf_maxw")
w = tf.ones((out_batch.shape[0]), dtype=tf.float32,name="w_ones") if gt_conf_pwr == 0:
else: w = tf.ones((out_batch.shape[0]), dtype=tf.float32,name="w_ones")
# 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: else:
w=tf.pow(w_clip, tf_gt_conf_pwr, name = "w") # 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")
if use_confidence:
tf_num_tilesf = tf.cast(tf_num_tiles, dtype=tf.float32, name="tf_num_tilesf") w_sub = tf.subtract (w_slice, tf_gt_conf_offset, name = "w_sub")
# conf_slice = tf.slice(out_batch,[0,1],[-1,1], name = "conf_slice") # w_clip = tf.clip_by_value(w_sub, tf_0f,tf_maxw, name = "w_clip")
conf_slice = tf.reshape(out_batch[:,1],[-1], name = "conf_slice") w_clip = tf.maximum(w_sub, tf_0f, name = "w_clip")
conf_sum = tf.reduce_sum(conf_slice, name = "conf_sum") if gt_conf_pwr == 1.0:
conf_avg = tf.divide(conf_sum, tf_num_tilesf, name = "conf_avg") w = w_clip
conf_avg1 = tf.subtract(conf_avg, tf_1f, name = "conf_avg1") else:
conf_avg2 = tf.square(conf_avg1, name = "conf_avg2") w=tf.pow(w_clip, tf_gt_conf_pwr, name = "w")
cost2 = tf.multiply (conf_avg2, tf_lambda_conf_avg, name = "cost2")
if use_confidence:
iconf_avg = tf.divide(tf_1f, conf_avg, name = "iconf_avg") tf_num_tilesf = tf.cast(tf_num_tiles, dtype=tf.float32, name="tf_num_tilesf")
nconf = tf.multiply (conf_slice, iconf_avg, name = "nconf") #normalized confidence # conf_slice = tf.slice(out_batch,[0,1],[-1,1], name = "conf_slice")
nconf_pwr = tf.pow(nconf, conf_pwr, name = "nconf_pwr") conf_slice = tf.reshape(out_batch[:,1],[-1], name = "conf_slice")
nconf_pwr_sum = tf.reduce_sum(nconf_pwr, name = "nconf_pwr_sum") conf_sum = tf.reduce_sum(conf_slice, name = "conf_sum")
nconf_pwr_offs = tf.subtract(nconf_pwr_sum, tf_1f, name = "nconf_pwr_offs") conf_avg = tf.divide(conf_sum, tf_num_tilesf, name = "conf_avg")
cost3 = tf.multiply (conf_avg2, nconf_pwr_offs, name = "cost3") conf_avg1 = tf.subtract(conf_avg, tf_1f, name = "conf_avg1")
w_all = tf.multiply (w, nconf, name = "w_all") conf_avg2 = tf.square(conf_avg1, name = "conf_avg2")
else: cost2 = tf.multiply (conf_avg2, tf_lambda_conf_avg, name = "cost2")
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") iconf_avg = tf.divide(tf_1f, conf_avg, name = "iconf_avg")
# d_gt_slice = tf.slice(gt_ds_batch,[0,0],[-1,1], name = "d_gt_slice") nconf = tf.multiply (conf_slice, iconf_avg, name = "nconf") #normalized confidence
disp_slice = tf.reshape(out_batch[:,0],[-1], name = "disp_slice") nconf_pwr = tf.pow(nconf, conf_pwr, name = "nconf_pwr")
d_gt_slice = tf.reshape(gt_ds_batch[:,0],[-1], name = "d_gt_slice") nconf_pwr_sum = tf.reduce_sum(nconf_pwr, name = "nconf_pwr_sum")
if absolute_disparity: nconf_pwr_offs = tf.subtract(nconf_pwr_sum, tf_1f, name = "nconf_pwr_offs")
out_diff = tf.subtract(disp_slice, d_gt_slice, name = "out_diff") cost3 = tf.multiply (conf_avg2, nconf_pwr_offs, name = "cost3")
else: w_all = tf.multiply (w, nconf, name = "w_all")
td_flat = tf.reshape(target_disparity_batch,[-1], name = "td_flat") else:
residual_disp = tf.subtract(d_gt_slice, td_flat, name = "residual_disp") w_all = w
out_diff = tf.subtract(disp_slice, residual_disp, name = "out_diff") # cost2 = 0.0
out_diff2 = tf.square(out_diff, name = "out_diff2") # cost3 = 0.0
out_wdiff2 = tf.multiply (out_diff2, w_norm, name = "out_wdiff2") # normalize weights
cost1 = tf.reduce_sum(out_wdiff2, name = "cost1") w_sum = tf.reduce_sum(w_all, name = "w_sum")
if use_confidence: iw_sum = tf.divide(tf_1f, w_sum, name = "iw_sum")
cost12 = tf.add(cost1, cost2, name = "cost12") w_norm = tf.multiply (w_all, iw_sum, name = "w_norm")
cost123 = tf.add(cost12, cost3, name = "cost123")
return cost123, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1 # disp_slice = tf.slice(out_batch,[0,0],[-1,1], name = "disp_slice")
else: # d_gt_slice = tf.slice(gt_ds_batch,[0,0],[-1,1], name = "d_gt_slice")
return cost1, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1 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")
if use_confidence:
cost12 = tf.add(cost1, 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 cost1, disp_slice, d_gt_slice, out_diff,out_diff2, w_norm, out_wdiff2, cost1
#corr2d325 = tf.concat([corr2d,target_disparity],0) #corr2d325 = tf.concat([corr2d,target_disparity],0)
......
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