diff --git a/data.py b/data.py
index 44420eea5d9b5e96585c94a9a50d418ee6b3979f..fc83570b23b74e3481db030321c09f5093a46d7b 100644
--- a/data.py
+++ b/data.py
@@ -22,7 +22,7 @@ class DataSet():
"""
self.seq_length = seq_length
self.class_limit = class_limit
- self.sequence_path = './data/sequences/'
+ self.sequence_path = os.path.join('data', 'sequences')
self.max_frames = 300 # max number of frames a video can have for us to use it
# Get the data.
@@ -39,7 +39,7 @@ class DataSet():
@staticmethod
def get_data():
"""Load our data from file."""
- with open('./data/data_file.csv', 'r') as fin:
+ with open(os.path.join('data', 'data_file.csv'), 'r') as fin:
reader = csv.reader(fin)
data = list(reader)
@@ -203,14 +203,14 @@ class DataSet():
def get_frames_for_sample(sample):
"""Given a sample row from the data file, get all the corresponding frame
filenames."""
- path = './data/' + sample[0] + '/' + sample[1] + '/'
+ path = os.path.join('data', sample[0], sample[1])
filename = sample[2]
- images = sorted(glob.glob(path + filename + '*jpg'))
+ images = sorted(glob.glob(os.path.join(path, filename + '*jpg')))
return images
@staticmethod
def get_filename_from_image(filename):
- parts = filename.split('/')
+ parts = filename.split(os.path.sep)
return parts[-1].replace('.jpg', '')
@staticmethod
diff --git a/data/1_move_files.py b/data/1_move_files.py
index 874ea1e22ccf3c3ebf8bf3cdf4114bfb6b361bb3..23f663f1c0028a675ff78aa57d23dc3c9e872f5a 100644
--- a/data/1_move_files.py
+++ b/data/1_move_files.py
@@ -13,8 +13,8 @@ def get_train_test_lists(version='01'):
breakdowns we'll later use to move everything.
"""
# Get our files based on version.
- test_file = './ucfTrainTestlist/testlist' + version + '.txt'
- train_file = './ucfTrainTestlist/trainlist' + version + '.txt'
+ test_file = os.path.join('ucfTrainTestlist', 'testlist' + version + '.txt')
+ train_file = os.path.join('ucfTrainTestlist', 'trainlist' + version + '.txt')
# Build the test list.
with open(test_file) as fin:
@@ -44,14 +44,14 @@ def move_files(file_groups):
for video in videos:
# Get the parts.
- parts = video.split('/')
+ parts = video.split(os.path.sep)
classname = parts[0]
filename = parts[1]
# Check if this class exists.
- if not os.path.exists(group + '/' + classname):
+ if not os.path.exists(os.path.join(group, classname)):
print("Creating folder for %s/%s" % (group, classname))
- os.makedirs(group + '/' + classname)
+ os.makedirs(os.path.join(group, classname))
# Check if we have already moved this file, or at least that it
# exists to move.
@@ -60,7 +60,7 @@ def move_files(file_groups):
continue
# Move it.
- dest = group + '/' + classname + '/' + filename
+ dest = os.path.join(group, classname, filename)
print("Moving %s to %s" % (filename, dest))
os.rename(filename, dest)
diff --git a/data/2_extract_files.py b/data/2_extract_files.py
index 47c13e7c87c5cfc47a3eb0f91f747ae22c3fc262..bf7c524785e1098570c3d5f750215d5365644ae6 100644
--- a/data/2_extract_files.py
+++ b/data/2_extract_files.py
@@ -25,13 +25,13 @@ def extract_files():
`ffmpeg -i video.mpg image-%04d.jpg`
"""
data_file = []
- folders = ['./train/', './test/']
+ folders = ['train', 'test']
for folder in folders:
- class_folders = glob.glob(folder + '*')
+ class_folders = glob.glob(os.path.join(folder, '*'))
for vid_class in class_folders:
- class_files = glob.glob(vid_class + '/*.avi')
+ class_files = glob.glob(os.path.join(vid_class, '*.avi'))
for video_path in class_files:
# Get the parts of the file.
@@ -43,10 +43,9 @@ def extract_files():
# the info.
if not check_already_extracted(video_parts):
# Now extract it.
- src = train_or_test + '/' + classname + '/' + \
- filename
- dest = train_or_test + '/' + classname + '/' + \
- filename_no_ext + '-%04d.jpg'
+ src = os.path.join(train_or_test, classname, filename)
+ dest = os.path.join(train_or_test, classname,
+ filename_no_ext + '-%04d.jpg')
call(["ffmpeg", "-i", src, dest])
# Now get how many frames it is.
@@ -66,25 +65,25 @@ def get_nb_frames_for_video(video_parts):
"""Given video parts of an (assumed) already extracted video, return
the number of frames that were extracted."""
train_or_test, classname, filename_no_ext, _ = video_parts
- generated_files = glob.glob(train_or_test + '/' + classname + '/' +
- filename_no_ext + '*.jpg')
+ generated_files = glob.glob(os.path.join(train_or_test, classname,
+ filename_no_ext + '*.jpg'))
return len(generated_files)
def get_video_parts(video_path):
"""Given a full path to a video, return its parts."""
- parts = video_path.split('/')
- filename = parts[3]
+ parts = video_path.split(os.path.sep)
+ filename = parts[2]
filename_no_ext = filename.split('.')[0]
- classname = parts[2]
- train_or_test = parts[1]
+ classname = parts[1]
+ train_or_test = parts[0]
return train_or_test, classname, filename_no_ext, filename
def check_already_extracted(video_parts):
"""Check to see if we created the -0001 frame of this file."""
train_or_test, classname, filename_no_ext, _ = video_parts
- return bool(os.path.exists(train_or_test + '/' + classname +
- '/' + filename_no_ext + '-0001.jpg'))
+ return bool(os.path.exists(os.path.join(train_or_test, classname,
+ filename_no_ext + '-0001.jpg')))
def main():
"""
diff --git a/extract_features.py b/extract_features.py
index e14bce46ef3f29b2af7a8f17edc56e5ce457f279..b424e84b1dbf09cb74fce3068da951b87d2e94c4 100644
--- a/extract_features.py
+++ b/extract_features.py
@@ -32,8 +32,8 @@ pbar = tqdm(total=len(data.data))
for video in data.data:
# Get the path to the sequence for this video.
- path = './data/sequences/' + video[2] + '-' + str(seq_length) + \
- '-features.txt'
+ path = os.path.join('data', 'sequences', video[2] + '-' + str(seq_length) + \
+ '-features.txt')
# Check if we already have it.
if os.path.isfile(path):
diff --git a/models.py b/models.py
index 5669f440771fcb160abf134d2c1e9726638575fb..c10f126b0c2e8affea8031918bb4b1a16a8806d6 100644
--- a/models.py
+++ b/models.py
@@ -47,7 +47,7 @@ class ResearchModels():
self.model = self.lstm()
elif model == 'lrcn':
print("Loading CNN-LSTM model.")
- self.input_shape = (seq_length, 150, 150, 3)
+ self.input_shape = (seq_length, 80, 80, 3)
self.model = self.lrcn()
elif model == 'mlp':
print("Loading simple MLP.")
@@ -66,7 +66,7 @@ class ResearchModels():
sys.exit()
# Now compile the network.
- optimizer = Adam(lr=1e-4, decay=1e-6)
+ optimizer = Adam(lr=1e-5, decay=1e-6)
self.model.compile(loss='categorical_crossentropy', optimizer=optimizer,
metrics=metrics)
diff --git a/train.py b/train.py
index b3cdb5462da1592afffc365c0e55227b5c46117e..0febe4b5c0deb1c00a5d358e91d41484993f02b8 100644
--- a/train.py
+++ b/train.py
@@ -5,31 +5,28 @@ from keras.callbacks import TensorBoard, ModelCheckpoint, EarlyStopping, CSVLogg
from models import ResearchModels
from data import DataSet
import time
+import os.path
def train(data_type, seq_length, model, saved_model=None,
concat=False, class_limit=None, image_shape=None,
- load_to_memory=False):
- # Set variables.
- nb_epoch = 1000000
- batch_size = 32
-
+ load_to_memory=False, batch_size=32, nb_epoch=100):
# Helper: Save the model.
checkpointer = ModelCheckpoint(
- filepath='./data/checkpoints/' + model + '-' + data_type + \
- '.{epoch:03d}-{val_loss:.3f}.hdf5',
+ filepath=os.path.join('data', 'checkpoints', model + '-' + data_type + \
+ '.{epoch:03d}-{val_loss:.3f}.hdf5'),
verbose=1,
save_best_only=True)
# Helper: TensorBoard
- tb = TensorBoard(log_dir='./data/logs')
+ tb = TensorBoard(log_dir=os.path.join('data', 'logs'))
# Helper: Stop when we stop learning.
- early_stopper = EarlyStopping(patience=100000)
+ early_stopper = EarlyStopping(patience=5)
# Helper: Save results.
timestamp = time.time()
- csv_logger = CSVLogger('./data/logs/' + model + '-' + 'training-' + \
- str(timestamp) + '.log')
+ csv_logger = CSVLogger(os.path.join('data', 'logs', model + '-' + 'training-' + \
+ str(timestamp) + '.log'))
# Get the data and process it.
if image_shape is None:
@@ -78,18 +75,20 @@ def train(data_type, seq_length, model, saved_model=None,
steps_per_epoch=steps_per_epoch,
epochs=nb_epoch,
verbose=1,
- callbacks=[tb, early_stopper, csv_logger],
+ callbacks=[tb, early_stopper, csv_logger, checkpointer],
validation_data=val_generator,
- validation_steps=10)
+ validation_steps=40)
def main():
"""These are the main training settings. Set each before running
this file."""
- model = 'conv_3d' # see `models.py` for more
+ model = 'lrcn' # see `models.py` for more
saved_model = None # None or weights file
- class_limit = 2 # int, can be 1-101 or None
+ class_limit = 10 # int, can be 1-101 or None
seq_length = 40
- load_to_memory = True # pre-load the sequences into memory
+ load_to_memory = False # pre-load the sequences into memory
+ batch_size = 32
+ nb_epoch = 1000
# Chose images or features and image shape based on network.
if model == 'conv_3d':
@@ -97,7 +96,7 @@ def main():
image_shape = (80, 80, 3)
elif model == 'lrcn':
data_type = 'images'
- image_shape = (150, 150, 3)
+ image_shape = (80, 80, 3)
else:
data_type = 'features'
image_shape = None
@@ -110,7 +109,7 @@ def main():
train(data_type, seq_length, model, saved_model=saved_model,
class_limit=class_limit, concat=concat, image_shape=image_shape,
- load_to_memory=load_to_memory)
+ load_to_memory=load_to_memory, batch_size=batch_size, nb_epoch=nb_epoch)
if __name__ == '__main__':
main()
diff --git a/train_cnn.py b/train_cnn.py
index 898fadde57397b64f5b8e69466e3c419e60f41ee..0fb81dee369e47301517cc6352e0ebc200134383 100644
--- a/train_cnn.py
+++ b/train_cnn.py
@@ -14,12 +14,13 @@ from keras.models import Model
from keras.layers import Dense, GlobalAveragePooling2D
from keras.callbacks import ModelCheckpoint, TensorBoard, EarlyStopping
from data import DataSet
+import os.path
data = DataSet()
# Helper: Save the model.
checkpointer = ModelCheckpoint(
- filepath='./data/checkpoints/inception.{epoch:03d}-{val_loss:.2f}.hdf5',
+ filepath=os.path.join('data', 'checkpoints', 'inception.{epoch:03d}-{val_loss:.2f}.hdf5',)
verbose=1,
save_best_only=True)
@@ -27,7 +28,7 @@ checkpointer = ModelCheckpoint(
early_stopper = EarlyStopping(patience=10)
# Helper: TensorBoard
-tensorboard = TensorBoard(log_dir='./data/logs/')
+tensorboard = TensorBoard(log_dir=os.path.join('data', 'logs'))
def get_generators():
train_datagen = ImageDataGenerator(
@@ -41,14 +42,14 @@ def get_generators():
test_datagen = ImageDataGenerator(rescale=1./255)
train_generator = train_datagen.flow_from_directory(
- './data/train/',
+ os.path.join('data', 'train'),
target_size=(299, 299),
batch_size=32,
classes=data.classes,
class_mode='categorical')
validation_generator = test_datagen.flow_from_directory(
- './data/test/',
+ os.path.join('data', 'test'),
target_size=(299, 299),
batch_size=32,
classes=data.classes,
diff --git a/validate_cnn.py b/validate_cnn.py
index 5cb7b3918dcf41948a1ebbfa2cf4d4ed47ff1cf0..7f30b40a96d038af411a7e79989e75a1c45b282f 100644
--- a/validate_cnn.py
+++ b/validate_cnn.py
@@ -5,6 +5,7 @@ import numpy as np
import operator
import random
import glob
+import os.path
from data import DataSet
from processor import process_image
from keras.models import load_model
@@ -15,7 +16,7 @@ def main(nb_images=5):
model = load_model('data/checkpoints/inception.057-1.16.hdf5')
# Get all our test images.
- images = glob.glob('./data/test/**/*.jpg')
+ images = glob.glob(os.path.join('data', 'test', '**', '*.jpg'))
for _ in range(nb_images):
print('-'*80)