def save_npy(sett):
f = open(os.path.join(FLAGS.meta_dir, sett) + '.txt', 'r')
image_names = []
labels = []
line = f.readline()
while line:
l = line.split(' ')
if len(l) > 2:
image_name = l[0] + ' ' + l[1]
label = l[2]
else:
image_name = l[0]
label = l[1]
# image_name, label = line.split(' ')
label = label[0:-1]
image_names.append(image_name)
labels.append(int(label))
line = f.readline()
f.close()
for i, img_name in enumerate(image_names):
full_path = os.path.join(FLAGS.data_dir, img_name)
print('processing ' + full_path, file=ff)
img = plt.imread(full_path)
n = 0
for patch in get_patches(img, FLAGS.max_patches,
FLAGS.patch_size):
n = n + 1
img_name0 = img_name.split('/')[-1]
np.save(
os.path.join(
temp_name, sett, meta[labels[i]] + '_' +
img_name0 + '_' + str(n)) + '.npy', {
'label': labels[i],
'patch': patch
})
def save_npy(sett):
f = open(os.path.join(FLAGS.meta_dir, sett) + '.txt', 'r')
image_names = []
labels = []
line = f.readline()
while line:
image_name, label = line.split(' ')
label = label[0:-1]
image_names.append(image_name + '-' + FLAGS.extra + '.' +
FLAGS.format)
labels.append(int(label))
line = f.readline()
f.close()
for i, img_name in enumerate(image_names):
full_path = os.path.join(FLAGS.data_dir, meta[labels[i]],
img_name)
print('processing ' + full_path, file=ff)
ff.flush()
img = plt.imread(full_path)
n = 0
for patch in get_patches(img, FLAGS.max_patches,
FLAGS.patch_size):
n = n + 1
np.save(
os.path.join(
temp_name, sett, meta[labels[i]] + '_' + img_name +
'_' + str(n)) + '.npy', {
'label': labels[i],
'patch': patch
})
def save_crop(sett):
f = open(os.path.join(FLAGS.meta_dir, sett) + '.txt', 'r')
image_names = []
labels = []
line = f.readline()
while line:
image_name, label = line.split(' ')
label = label[0:-1]
image_names.append(image_name + '-' + extra + '.' + format)
labels.append(int(label))
line = f.readline()
f.close()
for i, img_name in enumerate(image_names):
full_path = os.path.join(data_dir, meta[labels[i]], img_name)
print('processing ' + full_path, file=ff)
ff.flush()
img = plt.imread(full_path)
for patch in get_patches(img, 1, 512):
plt.imsave(
os.path.join(
FLAGS.out_dir, meta[labels[i]] + '_' +
img_name.split('.')[0]) + '.png', patch)
def main(_):
tf.gfile.MakeDirs(os.path.join(FLAGS.out_dir, 'train'))
tf.gfile.MakeDirs(os.path.join(FLAGS.out_dir, 'valid'))
ff = open(FLAGS.out_file, 'w')
if not ff:
raise RuntimeError('OUTPUT FILE OPEN ERROR!!!!!!')
temp_name = '%.06f' % time.time()
if not tf.gfile.Exists(temp_name):
tf.gfile.MakeDirs(os.path.join(temp_name, 'train'))
tf.gfile.MakeDirs(os.path.join(temp_name, 'valid'))
train_name = os.path.join(FLAGS.out_dir, 'train', 'spc_train.tfrecords')
valid_name = os.path.join(FLAGS.out_dir, 'valid', 'spc_valid.tfrecords')
train_writer = tf.python_io.TFRecordWriter(train_name)
valid_writer = tf.python_io.TFRecordWriter(valid_name)
if not tf.gfile.Exists(FLAGS.meta_dir):
tf.gfile.MakeDirs(FLAGS.meta_dir)
classes = os.listdir(FLAGS.data_dir)
spc_classes = open(os.path.join(FLAGS.meta_dir, 'spc_classes.txt'),
'w')
train_list = open(os.path.join(FLAGS.meta_dir, 'train.txt'), 'w')
valid_list = open(os.path.join(FLAGS.meta_dir, 'valid.txt'), 'w')
for label, class_name in enumerate(classes):
spc_classes.write(('%d ' % label) + class_name + '\n')
scenes = os.listdir(os.path.join(FLAGS.data_dir, class_name))
dice = np.random.randint(0, 5, 1)
a = dice
for j, scene in enumerate(scenes):
sett = 'valid' if j == a else 'train'
img_names = os.listdir(
os.path.join(FLAGS.data_dir, class_name, scene))
for img_name in img_names:
full_path = os.path.join(FLAGS.data_dir, class_name, scene,
img_name)
if sett == 'train':
train_list.write(
os.path.join(class_name, scene, img_name) +
(' %d\n' % label))
else:
valid_list.write(
os.path.join(class_name, scene, img_name) +
(' %d\n' % label))
print('processing ' + full_path, file=ff)
img = plt.imread(full_path)
dice = np.random.randint(0, 5, 1)
n = 0
for patch in get_patches(img, FLAGS.max_patches,
FLAGS.patch_size):
n = n + 1
np.save(
os.path.join(
temp_name, sett, class_name + '_' + img_name +
'_' + str(n)) + '.npy', {
'label': label,
'patch': patch
})
spc_classes.close()
train_list.close()
valid_list.close()
else:
f = open(os.path.join(FLAGS.meta_dir, 'spc_classes.txt'), 'r')
meta = {}
line = f.readline()
while line:
label, class_name = line.split(' ')
class_name = class_name[0:-1]
meta[int(label)] = class_name
line = f.readline()
f.close()
def save_npy(sett):
f = open(os.path.join(FLAGS.meta_dir, sett) + '.txt', 'r')
image_names = []
labels = []
line = f.readline()
while line:
l = line.split(' ')
if len(l) > 2:
image_name = l[0] + ' ' + l[1]
label = l[2]
else:
image_name = l[0]
label = l[1]
# image_name, label = line.split(' ')
label = label[0:-1]
image_names.append(image_name)
labels.append(int(label))
line = f.readline()
f.close()
for i, img_name in enumerate(image_names):
full_path = os.path.join(FLAGS.data_dir, img_name)
print('processing ' + full_path, file=ff)
img = plt.imread(full_path)
n = 0
for patch in get_patches(img, FLAGS.max_patches,
FLAGS.patch_size):
n = n + 1
img_name0 = img_name.split('/')[-1]
np.save(
os.path.join(
temp_name, sett, meta[labels[i]] + '_' +
img_name0 + '_' + str(n)) + '.npy', {
'label': labels[i],
'patch': patch
})
save_npy('train')
save_npy('valid')
train = os.listdir(os.path.join(temp_name, 'train'))
valid = os.listdir(os.path.join(temp_name, 'valid'))
print(len(train))
print(len(valid))
idx = list(range(len(train)))
shuffle(idx)
for i in idx:
dic = np.load(os.path.join(temp_name, 'train', train[i])).item()
patch = dic['patch']
label = dic['label']
patch_raw = patch.tostring()
label_raw = np.array([label]).astype(np.int32).tostring()
example = tf.train.Example(features=tf.train.Features(
feature={
'patch_raw': _bytes_feature(patch_raw),
'label': _bytes_feature(label_raw)
}))
train_writer.write(example.SerializeToString())
idx = list(range(len(valid)))
shuffle(idx)
for i in idx:
dic = np.load(os.path.join(temp_name, 'valid', valid[i])).item()
patch = dic['patch']
label = dic['label']
patch_raw = patch.tostring()
label_raw = np.array([label]).astype(np.int32).tostring()
example = tf.train.Example(features=tf.train.Features(
feature={
'patch_raw': _bytes_feature(patch_raw),
'label': _bytes_feature(label_raw)
}))
valid_writer.write(example.SerializeToString())
train_writer.close()
valid_writer.close()
ff.close()
tf.gfile.DeleteRecursively(temp_name)
def main(_):
ff = open(FLAGS.out_file, 'w')
if not ff:
raise RuntimeError('OUTPUT FILE OPEN ERROR!!!!!!')
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS.gpu
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.7
config.gpu_options.allow_growth = True
if not tf.gfile.Exists(FLAGS.data_dir):
raise RuntimeError('data direction is not exist!')
with tf.name_scope('input'):
x = tf.placeholder(tf.float32, [None, FLAGS.patch_size, FLAGS.patch_size, 3], 'x')
y = build.net(x, False, FLAGS)
# update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
# with tf.control_dependencies(update_ops):
pred = tf.nn.softmax(y, 1)
with tf.name_scope("saver"):
saver = tf.train.Saver(name="saver")
f = open(os.path.join(FLAGS.meta_dir, FLAGS.set) + '.txt', 'r')
image_names = []
labels = []
line = f.readline()
while line:
l = line.split(' ')
if len(l) == 2:
image_name = l[0]
label = l[1]
else:
image_name = l[0] + ' ' + l[1]
label = l[2]
# image_name, label = line.split(' ')
label = label[0:-1]
image_names.append(image_name.split('.')[0] + '-' + FLAGS.extra + '.' + FLAGS.format)
labels.append(int(label))
line = f.readline()
f.close()
f = open(os.path.join(FLAGS.meta_dir, 'spc_classes.txt'), 'r')
meta = {}
line = f.readline()
while line:
label, class_name = line.split(' ')
class_name = class_name[0:-1]
meta[int(label)] = class_name
line = f.readline()
f.close()
confusion = np.zeros(shape=(10, 10), dtype=np.uint32)
confusion_i = np.zeros(shape=(10, 10), dtype=np.uint32)
total = 0.
correct = 0.
total_p = 0.
correct_p = 0.
with tf.Session(config = config) as sess:
if tf.gfile.Exists(os.path.join(FLAGS.ckpt_dir, 'checkpoint')):
saver.restore(sess, tf.train.latest_checkpoint(FLAGS.ckpt_dir) if FLAGS.model_name is None else os.path.join(FLAGS.ckpt_dir, FLAGS.model_name))
else:
raise RuntimeError("Check point files don't exist!")
for i in range(len(labels)):
label = labels[i]
class_name = meta[label]
image_name = image_names[i]
full_path = os.path.join(FLAGS.data_dir, class_name, image_name)
img = plt.imread(full_path)
for img in get_patches(img, 1, 512):
data = np.ndarray(shape=(FLAGS.patches, FLAGS.patch_size, FLAGS.patch_size, 3), dtype=np.float32)
for n, patch in enumerate(get_patches(img, FLAGS.patches, FLAGS.patch_size)):
patch = standardization(patch)
data[n, :] = patch
# data = standardization(data)
prediction = sess.run(pred, feed_dict={x: data})
prediction0 = np.argmax(prediction, 1)
for n in prediction0:
if n == label:
correct_p = correct_p + 1
confusion[label, n] = confusion[label, n] + 1
total_p = total_p + FLAGS.patches
# count = np.bincount(prediction)
# prediction = np.argmax(count)
prediction = np.sum(prediction, 0)
#print(prediction)
prediction = np.argmax(prediction)
confusion_i[label, prediction] = confusion_i[label, prediction] + 1
print("predict %d while true label is %d." % (prediction, label), file=ff)
ff.flush()
total = total + 1
if prediction == label:
correct = correct + 1
print('accuracy(patch level) = %f' % (correct_p / total_p), file=ff)
print('accuracy(image level) = %f' % (correct / total), file=ff)
print('confusion matrix--patch level:', file=ff)
print(confusion, file=ff)
print('confusion matrix--image level:', file=ff)
print(confusion_i, file=ff)
print('/|\\', file=ff)
print(' |', file=ff)
print('actual', file=ff)
print(' |', file=ff)
print(' ---prediction--->', file=ff)
ff.close()
def main(_):
ff = open(FLAGS.out_file, 'w')
if not ff:
raise RuntimeError('OUTPUT FILE OPEN ERROR!!!!!!')
print('fname,camera', file=ff)
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS.gpu
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.7
config.gpu_options.allow_growth = True
with tf.name_scope('input'):
x = tf.placeholder(tf.float32,
[None, FLAGS.patch_size, FLAGS.patch_size, 3], 'x')
y = build.net(x, False, FLAGS)
# update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
# with tf.control_dependencies(update_ops):
pred = tf.nn.softmax(y, 1)
if FLAGS.ema:
variable_averages = tf.train.ExponentialMovingAverage(0.9999)
variables_to_restore = variable_averages.variables_to_restore()
saver = tf.train.Saver(variables_to_restore, name='saver')
else:
saver = tf.train.Saver(name="saver")
# f = open(os.path.join(FLAGS.meta_dir, FLAGS.set) + '.txt', 'r')
# image_names = []
# labels = []
# line = f.readline()
# while line:
# l = line.split(' ')
# if len(l) == 2:
# image_name = l[0]
# label = l[1]
# else:
# image_name = l[0] + ' ' + l[1]
# label = l[2]
# # image_name, label = line.split(' ')
# label = label[0:-1]
# image_names.append(image_name.split('.')[0] + '-' + FLAGS.extra + '.' + FLAGS.format)
# labels.append(int(label))
# line = f.readline()
# f.close()
image_names = os.listdir(FLAGS.data_dir)
f = open(os.path.join(FLAGS.meta_dir, 'spc_classes.txt'), 'r')
meta = {}
line = f.readline()
while line:
label, class_name = line.split(' ')
class_name = class_name[0:-1]
meta[int(label)] = class_name
line = f.readline()
f.close()
# confusion = np.zeros(shape=(10, 10), dtype=np.uint32)
# confusion_i = np.zeros(shape=(10, 10), dtype=np.uint32)
# total = 0.
# correct = 0.
# total_p = 0.
# correct_p = 0.
with tf.Session(config=config) as sess:
if tf.gfile.Exists(os.path.join(FLAGS.ckpt_dir, 'checkpoint')):
saver.restore(
sess,
tf.train.latest_checkpoint(FLAGS.ckpt_dir)
if FLAGS.model_name is None else os.path.join(
FLAGS.ckpt_dir, FLAGS.model_name))
else:
raise RuntimeError("Check point files don't exist!")
for i in range(len(image_names)):
# label = labels[i]
# class_name = meta[label]
image_name = image_names[i]
full_path = os.path.join(FLAGS.data_dir, image_name)
img = plt.imread(full_path)
if img.shape[2] == 4:
img = img[:, :, 0:3]
data = np.ndarray(shape=(FLAGS.patches, FLAGS.patch_size,
FLAGS.patch_size, 3),
dtype=np.float32)
for n, patch in enumerate(
get_patches(img, FLAGS.patches, FLAGS.patch_size)):
patch = standardization(patch)
data[n, :] = patch
# data = standardization(data)
prediction = sess.run(pred, feed_dict={x: data})
prediction = np.argmax(prediction, 1)
# for n in prediction0:
# if n == label:
# correct_p = correct_p + 1
# confusion[label, n] = confusion[label, n] + 1
# total_p = total_p + FLAGS.patches
count = np.bincount(prediction)
prediction = np.argmax(count)
# prediction = np.sum(prediction, 0)
#print(prediction)
# prediction = np.argmax(prediction)
print("%s,%s" % (image_name, meta[prediction]), file=ff)
ff.flush()
ff.close()
def main(_):
# ff = open(FLAGS.out_file, 'w')
# if not ff:
# raise RuntimeError('OUTPUT FILE OPEN ERROR!!!!!!')
# print('fname,camera', file=ff)
os.environ["CUDA_VISIBLE_DEVICES"] = FLAGS.gpu
config = tf.ConfigProto()
config.gpu_options.per_process_gpu_memory_fraction = 0.7
config.gpu_options.allow_growth = True
with tf.name_scope('input'):
x = tf.placeholder(tf.float32, [None, FLAGS.patch_size, FLAGS.patch_size, 3], 'x')
w, _ = build_net(x, False, FLAGS)
# update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
# with tf.control_dependencies(update_ops):
# pred = tf.nn.softmax(y, 1)
with tf.name_scope("saver"):
saver = tf.train.Saver(name="saver")
# f = open(os.path.join(FLAGS.meta_dir, FLAGS.set) + '.txt', 'r')
# image_names = []
# labels = []
# line = f.readline()
# while line:
# l = line.split(' ')
# if len(l) == 2:
# image_name = l[0]
# label = l[1]
# else:
# image_name = l[0] + ' ' + l[1]
# label = l[2]
# # image_name, label = line.split(' ')
# label = label[0:-1]
# image_names.append(image_name.split('.')[0] + '-' + FLAGS.extra + '.' + FLAGS.format)
# labels.append(int(label))
# line = f.readline()
# f.close()
image_names = os.listdir('/data/spcup_test')
# f = open(os.path.join(FLAGS.meta_dir, 'spc_classes.txt'), 'r')
# meta = {}
# line = f.readline()
# while line:
# label, class_name = line.split(' ')
# class_name = class_name[0:-1]
# meta[int(label)] = class_name
# line = f.readline()
# f.close()
# confusion = np.zeros(shape=(10, 10), dtype=np.uint32)
# confusion_i = np.zeros(shape=(10, 10), dtype=np.uint32)
# total = 0.
# correct = 0.
# total_p = 0.
# correct_p = 0.
with tf.Session(config = config) as sess:
if tf.gfile.Exists(os.path.join(FLAGS.ckpt_dir, 'checkpoint')):
saver.restore(sess, tf.train.latest_checkpoint(FLAGS.ckpt_dir) if FLAGS.model_name is None else os.path.join(FLAGS.ckpt_dir, FLAGS.model_name))
else:
raise RuntimeError("Check point files don't exist!")
for i in range(len(image_names)):
# label = labels[i]
# class_name = meta[label]
image_name = image_names[i]
print(image_name)
full_path = os.path.join('/data/spcup_test', image_name)
img = plt.imread(full_path)
data = np.ndarray(shape=(FLAGS.patches, FLAGS.patch_size, FLAGS.patch_size, 3), dtype=np.float32)
for n, patch in enumerate(get_patches(img, FLAGS.patches, FLAGS.patch_size)):
patch = standardization(patch)
data[n, :] = patch
# data = standardization(data)
ww = sess.run(w, feed_dict={x: data})[0, :]
# print(ww.shape)
plt.imshow(ww)
plt.show()