def main(args):
dataset_json = args.json
tuning_layers = []
if args.tune:
tuning_layers = args.tune.split(",")
print tuning_layers
vgg_path = args.vgg
max_frames = args.max_frames
print "Loading VGGNet model. Model type: %s" % args.model
if args.model == 'average':
model = AverageFrameModel(vgg_path,
output_neurons=4,
tuning_layers=tuning_layers)
elif args.model == 'late':
model = LateFusionModel(vgg_path, output_neurons=4)
else:
raise ValueError("Model type must be one of 'average' and 'late'")
print "Reading data"
start = datetime.datetime.now()
tvt_split = [args.train, args.val, args.test]
data = read_dataset_tvt(dataset_json,
sample_probability=0.5,
mode='temporal',
max_frames=max_frames,
mean_value=model.mean_bgr,
tvt_split=tvt_split,
ids_file=args.ids)
print "Training data shape:", data["train_X"].shape
print "Test data shape:", data["test_X"].shape
print "Validation data shape:", data["val_X"].shape
end = datetime.datetime.now()
print "Read data in %d seconds" % (end - start).seconds
print "---- Training parameters summary -----"
param_summary(data["train_X"].shape[0], args)
batch_size = min(args.batch_size, data["train_X"].shape[0])
print "--------------------------------------"
solver = Solver(model,
data["train_X"],
data["train_y"],
val_X=data["val_X"],
val_y=data["val_y"],
model_type=args.model,
num_epochs=args.num_epochs,
batch_size=batch_size,
output_lr=args.output_lr,
tune_lr=args.tune_lr,
tuning_layers=tuning_layers,
reg=args.reg)
solver.train()
test_predictions, scores = solver.predict(data["test_X"], data["test_y"])
write_predictions(args.model, args.ids, test_predictions, scores, args.out)
print "--------------------------------------"
print "---- Training parameters summary -----"
param_summary(data["train_X"].shape[0], args)
print "---- Loss and accuracy history ----"
print "Training loss history"
print solver.train_loss_history
print "Training accuracy history"
print solver.train_acc_history
print "Validation accuracy history"
print solver.val_acc_history
class ClassifierEngine(object):
def __init__(self, method='fcn', configs=None):
self.model = None
if method is 'fcn':
self.model = FullyConnectedNet(configs['arch'])
elif method is 'cnn':
self.model = ConvNet(configs['arch'])
else:
raise ValueError('Unrecognized classifier ...')
self.solver = Solver(self.model, configs['solver'])
def show_history(self):
''' show the history of training process '''
plt.rcParams['figure.figsize'] = (10.0, 8.0)
plt.rcParams['image.interpolation'] = 'nearest'
plt.rcParams['image.cmap'] = 'gray'
plt.subplot(2, 1, 1)
plt.plot(self.solver.loss_history)
plt.xlabel('Iteration')
plt.ylabel('Loss')
plt.title('Loss history')
plt.subplot(2, 1, 2)
plt.plot(self.solver.train_acc_history, label='train')
plt.plot(self.solver.val_acc_history, label='val')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.title('Classification accuracy history')
fig_version = time.strftime("%Y%m%d%H%M%S",
time.localtime(int(time.time())))
plt.savefig(
'../dataset/training_details/loss_curve/training_details_%s.png' %
fig_version,
dpi=400)
# plt.show();
plt.close()
def visualize_filters(self):
''' visualize the filters '''
grid = visualize_grid(self.model.params['W1'].transpose(0, 2, 3, 1))
plt.imshow(grid.astype('uint8'))
plt.axis('off')
plt.gcf().set_size_inches(5, 5)
fig_version = time.strftime("%Y%m%d%H%M%S",
time.localtime(int(time.time())))
plt.savefig('../dataset/training_details/filters/filters_%s.png' %
fig_version,
dpi=400)
# plt.show();
plt.close()
def classifier_runner(self, verbose=True):
'''
step - 1. train a fcn_classifier
step - 2. show the training-curve [-optional-]
step - 3. predict the test_date
'''
self.solver.train()
self.solver.predict()
if verbose:
self.show_history()
self.visualize_filters()
def main(args):
dataset_json = args.json
tuning_layers = []
if args.tune:
tuning_layers = args.tune.split(",")
print tuning_layers
vgg_path = args.vgg
max_frames = args.max_frames
print "Loading VGGNet model. Model type: %s" % args.model
if args.model == 'average':
model = AverageFrameModel(vgg_path,
output_neurons=4,
tuning_layers=tuning_layers)
elif args.model == 'late':
model = LateFusionModel(vgg_path, output_neurons=4)
else:
raise ValueError("Model type must be one of 'average' and 'late'")
print "Reading data"
start = datetime.datetime.now()
tvt_split = [args.train, args.val, args.test]
data = read_dataset_tvt(dataset_json, sample_probability=0.5, mode='temporal', max_frames=max_frames,
mean_value=model.mean_bgr, tvt_split=tvt_split, ids_file=args.ids)
print "Training data shape:", data["train_X"].shape
print "Test data shape:", data["test_X"].shape
print "Validation data shape:", data["val_X"].shape
end = datetime.datetime.now()
print "Read data in %d seconds" % (end-start).seconds
print "---- Training parameters summary -----"
param_summary(data["train_X"].shape[0], args)
batch_size = min(args.batch_size, data["train_X"].shape[0])
print "--------------------------------------"
solver = Solver(model,
data["train_X"], data["train_y"],
val_X=data["val_X"], val_y=data["val_y"],
model_type=args.model,
num_epochs=args.num_epochs,
batch_size=batch_size,
output_lr=args.output_lr,
tune_lr=args.tune_lr,
tuning_layers=tuning_layers,
reg=args.reg)
solver.train()
test_predictions, scores = solver.predict(data["test_X"], data["test_y"])
write_predictions(args.model, args.ids, test_predictions, scores, args.out)
print "--------------------------------------"
print "---- Training parameters summary -----"
param_summary(data["train_X"].shape[0], args)
print "---- Loss and accuracy history ----"
print "Training loss history"
print solver.train_loss_history
print "Training accuracy history"
print solver.train_acc_history
print "Validation accuracy history"
print solver.val_acc_history