diff --git a/bob/learn/tensorflow/datashuffler/TripletWithSelectionDisk.py b/bob/learn/tensorflow/datashuffler/TripletWithSelectionDisk.py
index a41deb9988d1bf3a460780a611e5bb4527a19e08..2a09b67cd26a0cf56f02d8308115e9377501993f 100644
--- a/bob/learn/tensorflow/datashuffler/TripletWithSelectionDisk.py
+++ b/bob/learn/tensorflow/datashuffler/TripletWithSelectionDisk.py
@@ -52,7 +52,6 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
def __init__(self, data, labels,
input_shape,
input_dtype="float64",
- scale=True,
batch_size=1,
seed=10,
data_augmentation=None,
@@ -64,7 +63,6 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
labels=labels,
input_shape=input_shape,
input_dtype=input_dtype,
- scale=scale,
batch_size=batch_size,
seed=seed,
data_augmentation=data_augmentation,
@@ -93,13 +91,9 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
for i in range(self.shape[0]):
file_name_a, file_name_p, file_name_n = self.get_one_triplet(self.data, self.labels)
- sample_a[i, ...] = self.load_from_file(str(file_name_a))
- sample_p[i, ...] = self.load_from_file(str(file_name_p))
- sample_n[i, ...] = self.load_from_file(str(file_name_n))
-
- sample_a = self.normalize_sample(sample_a)
- sample_p = self.normalize_sample(sample_p)
- sample_n = self.normalize_sample(sample_n)
+ sample_a[i, ...] = self.normalize_sample(self.load_from_file(str(file_name_a)))
+ sample_p[i, ...] = self.normalize_sample(self.load_from_file(str(file_name_p)))
+ sample_n[i, ...] = self.normalize_sample(self.load_from_file(str(file_name_n)))
return [sample_a, sample_p, sample_n]
@@ -181,7 +175,6 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
The best positive sample for the anchor is the farthest from the anchor
"""
- #logger.info("****************** numpy.where")
indexes = numpy.where(self.labels == label)[0]
numpy.random.shuffle(indexes)
@@ -190,26 +183,19 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
distances = []
shape = tuple([len(indexes)] + list(self.shape[1:]))
sample_p = numpy.zeros(shape=shape, dtype='float32')
- #logger.info("****************** search")
+
for i in range(shape[0]):
- #logger.info("****************** fetch")
file_name = self.data[indexes[i], ...]
- #logger.info("****************** load")
- sample_p[i, ...] = self.load_from_file(str(file_name))
+ sample_p[i, ...] = self.normalize_sample(self.load_from_file(str(file_name)))
- sample_p = self.normalize_sample(sample_p)
-
- #logger.info("****************** project")
embedding_p = self.project(sample_p)
- #logger.info("****************** distances")
# Projecting the positive instances
for p in embedding_p:
distances.append(euclidean(embedding_a, p))
# Geting the max
index = numpy.argmax(distances)
- #logger.info("****************** return")
return sample_p[index, ...], distances[index]
def get_negative(self, label, embedding_a, distance_anchor_positive):
@@ -220,7 +206,6 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
#anchor_feature = self.feature_extractor(self.reshape_for_deploy(anchor), session=self.session)
# Selecting the negative samples
- #logger.info("****************** numpy.where")
indexes = numpy.where(self.labels != label)[0]
numpy.random.shuffle(indexes)
indexes = indexes[
@@ -228,20 +213,13 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
shape = tuple([len(indexes)] + list(self.shape[1:]))
sample_n = numpy.zeros(shape=shape, dtype='float32')
- #logger.info("****************** search")
for i in range(shape[0]):
- #logger.info("****************** fetch")
file_name = self.data[indexes[i], ...]
- #logger.info("****************** load")
- sample_n[i, ...] = self.load_from_file(str(file_name))
-
- sample_n = self.normalize_sample(sample_n)
+ sample_n[i, ...] = self.normalize_sample(self.load_from_file(str(file_name)))
- #logger.info("****************** project")
embedding_n = self.project(sample_n)
distances = []
- #logger.info("****************** distances")
for n in embedding_n:
d = euclidean(embedding_a, n)
@@ -258,5 +236,4 @@ class TripletWithSelectionDisk(Triplet, Disk, OnLineSampling):
if numpy.isinf(distances[index]):
logger.info("SEMI-HARD negative not found, took the first one")
index = 0
- #logger.info("****************** return")
return sample_n[index, ...]
diff --git a/bob/learn/tensorflow/test/test_datashuffler.py b/bob/learn/tensorflow/test/test_datashuffler.py
index 648fdbedaa8dca3976b20d40ab38c84400b6e1ca..60abbc951f6b9d6f5daa805b69c9e49d84f53436 100644
--- a/bob/learn/tensorflow/test/test_datashuffler.py
+++ b/bob/learn/tensorflow/test/test_datashuffler.py
@@ -4,7 +4,8 @@
# @date: Thu 13 Oct 2016 13:35 CEST
import numpy
-from bob.learn.tensorflow.datashuffler import Memory, SiameseMemory, TripletMemory, Disk, SiameseDisk, TripletDisk
+from bob.learn.tensorflow.datashuffler import Memory, SiameseMemory, TripletMemory, Disk, SiameseDisk, TripletDisk, \
+ TripletWithFastSelectionDisk, TripletWithSelectionDisk
import pkg_resources
from bob.learn.tensorflow.utils import load_mnist
import os
@@ -15,7 +16,6 @@ Some unit tests for the datashuffler
def get_dummy_files():
-
base_path = pkg_resources.resource_filename(__name__, 'data/dummy_database')
files = []
clients = []
@@ -28,7 +28,6 @@ def get_dummy_files():
def test_memory_shuffler():
-
train_data, train_labels, validation_data, validation_labels = load_mnist()
train_data = numpy.reshape(train_data, (train_data.shape[0], 28, 28, 1))
@@ -50,7 +49,6 @@ def test_memory_shuffler():
def test_siamesememory_shuffler():
-
train_data, train_labels, validation_data, validation_labels = load_mnist()
train_data = numpy.reshape(train_data, (train_data.shape[0], 28, 28, 1))
@@ -74,7 +72,6 @@ def test_siamesememory_shuffler():
def test_tripletmemory_shuffler():
-
train_data, train_labels, validation_data, validation_labels = load_mnist()
train_data = numpy.reshape(train_data, (train_data.shape[0], 28, 28, 1))
@@ -98,7 +95,6 @@ def test_tripletmemory_shuffler():
def test_disk_shuffler():
-
train_data, train_labels = get_dummy_files()
batch_shape = [2, 125, 125, 3]
@@ -119,7 +115,6 @@ def test_disk_shuffler():
def test_siamesedisk_shuffler():
-
train_data, train_labels = get_dummy_files()
batch_shape = [2, 125, 125, 3]
@@ -142,7 +137,6 @@ def test_siamesedisk_shuffler():
def test_tripletdisk_shuffler():
-
train_data, train_labels = get_dummy_files()
batch_shape = [1, 125, 125, 3]
@@ -164,3 +158,47 @@ def test_tripletdisk_shuffler():
assert placeholders[2].get_shape().as_list() == batch_shape
+def test_triplet_fast_selection_disk_shuffler():
+ train_data, train_labels = get_dummy_files()
+
+ batch_shape = [1, 125, 125, 3]
+
+ data_shuffler = TripletWithFastSelectionDisk(train_data, train_labels,
+ input_shape=batch_shape[1:],
+ total_identities=1,
+ batch_size=batch_shape[0])
+
+ batch = data_shuffler.get_batch()
+
+ assert len(batch) == 3
+ assert batch[0].shape == tuple(batch_shape)
+ assert batch[1].shape == tuple(batch_shape)
+ assert batch[2].shape == tuple(batch_shape)
+
+ placeholders = data_shuffler.get_placeholders(name="train")
+ assert placeholders[0].get_shape().as_list() == batch_shape
+ assert placeholders[1].get_shape().as_list() == batch_shape
+ assert placeholders[2].get_shape().as_list() == batch_shape
+
+
+def test_triplet_selection_disk_shuffler():
+ train_data, train_labels = get_dummy_files()
+
+ batch_shape = [1, 125, 125, 3]
+
+ data_shuffler = TripletWithSelectionDisk(train_data, train_labels,
+ input_shape=batch_shape[1:],
+ total_identities=1,
+ batch_size=batch_shape[0])
+
+ batch = data_shuffler.get_batch()
+
+ assert len(batch) == 3
+ assert batch[0].shape == tuple(batch_shape)
+ assert batch[1].shape == tuple(batch_shape)
+ assert batch[2].shape == tuple(batch_shape)
+
+ placeholders = data_shuffler.get_placeholders(name="train")
+ assert placeholders[0].get_shape().as_list() == batch_shape
+ assert placeholders[1].get_shape().as_list() == batch_shape
+ assert placeholders[2].get_shape().as_list() == batch_shape