def eval(config, solver, epoch=0):
acc = 0
loss = 0
labels = []
predictions = []
test_file = os.path.join(config.data, 'testrotnet.txt')
mean = get_mean(config.mean_file)
batch_size = solver.test_nets[0].blobs['data'].num
test_iters = int(
get_dataset_size(config, 'testrotnet') / config.batch_size)
for i in range(test_iters):
solver.test_nets[0].forward()
acc += solver.test_nets[0].blobs['my_accuracy'].data
loss += solver.test_nets[0].blobs['(automatic)'].data
probs = solver.test_nets[0].blobs['prob'].data
predictions += classify(probs)
labels.append(int(solver.test_nets[0].blobs['label'].data[0]))
acc /= test_iters
loss /= test_iters
if not config.test:
print("Accuracy: {:.3f}".format(acc))
print("Loss: {:.3f}".format(loss))
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
else:
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def eval_alone(config):
data = model_data.read_data(config.data, config, read_train=False)
data = data.test
seq_rnn_model = SequenceRNNModel(config.n_input_fc, config.num_views, config.n_hidden, config.decoder_embedding_size, config.num_classes+1, config.n_hidden,
batch_size=data.size(),
is_training=False,
use_lstm=config.use_lstm,
use_attention=config.use_attention,
use_embedding=config.use_embedding, num_heads=config.num_heads)
seq_rnn_model.build_model("train")
tf_config = tf.ConfigProto()
tf_config.gpu_options.allow_growth = True
with tf.Session(config=tf_config) as sess:
#config.gpu_options.per_process_gpu_memory_fraction = 0.3
saver = tf.train.Saver()
saver.restore(sess, get_modelpath(config.weights))
acc, loss, predictions, labels = _test(data, seq_rnn_model, sess)
log(config.log_file, "TESTING ACCURACY {}".format(acc))
predictions = [x-1 for x in predictions]
labels = [x-1 for x in labels]
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels, config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def eval(config, solver, epoch=0):
acc = 0
loss = 0
all_labels = []
predictions = []
test_count = get_dataset_size(config, 'test')
keys = solver.test_nets[0].blobs.keys()
batch_size = (solver.test_nets[0].blobs['label_octreedatabase_1_split_0'].
data.shape[0])
test_iters = test_count / batch_size
logits = np.zeros((test_count, config.num_classes))
print(logits.shape)
for i in range(test_iters):
solver.test_nets[0].forward()
loss += solver.test_nets[0].blobs['loss'].data
probs = solver.test_nets[0].blobs['ip2'].data
logits[i * batch_size:(i + 1) * batch_size] = probs
all_labels += list(solver.test_nets[0].blobs['label'].data)
solver.test_nets[0].forward()
loss += solver.test_nets[0].blobs['loss'].data
probs = solver.test_nets[0].blobs['ip2'].data
logits[batch_size * (test_iters):] = probs[0:test_count % batch_size]
all_labels += list(solver.test_nets[0].blobs['label'].data)[0:test_count %
batch_size]
loss /= test_iters + 1
predictions = []
labels = []
for i in range(len(all_labels) / config.num_rotations):
predictions.append(
np.argmax(
np.sum(logits[i * config.num_rotations:(i + 1) *
config.num_rotations],
axis=0)))
labels.append(all_labels[i * config.num_rotations])
acc = sum([1 for i in range(len(labels)) if predictions[i] == labels[i]
]) / float(len(labels))
if not config.test:
log(config.log_file, "EPOCH: {} Test loss: {}".format(epoch, loss))
log(config.log_file, "EPOCH: {} Test accuracy: {}".format(epoch, acc))
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
else:
log(config.log_file, "----------------------")
import Evaluation_tools as et
labels = [int(l) for l in labels]
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def test(config,test_vertices, test_faces, test_nFaces, test_labels):
log(config['log_file'],"Start testing")
config['mode'] = 'test'
_, predictions = acc_fun(net,test_vertices, test_faces, test_nFaces, test_labels, config)
acc = 100.*(predictions == test_labels).sum()/len(test_labels)
log(config['log_file'],'Eval accuracy: {}'.format(acc))
import Evaluation_tools as et
eval_file = os.path.join(config['log_dir'], '{}.txt'.format(config['name']))
print(eval_file)
et.write_eval_file(config['data'], eval_file, predictions , test_labels , config['name'])
et.make_matrix(config['data'], eval_file, config['log_dir'])
def test(model, config, best_accuracy=0, epoch=None):
batch_amount = 0
model.test_loss.data.zero_()
model.test_accuracy.data.zero_()
predictions = []
labels = []
for i, data in enumerate(testloader):
input_pc, input_sn, input_label, input_node, input_node_knn_I = data
model.set_input(input_pc, input_sn, input_label, input_node,
input_node_knn_I)
model.test_model()
batch_amount += input_label.size()[0]
model.test_loss += model.loss.detach() * input_label.size()[0]
# accumulate accuracy
_, predicted_idx = torch.max(model.score.data, dim=1, keepdim=False)
predictions += list(predicted_idx)
labels += list(input_label)
correct_mask = torch.eq(predicted_idx, model.input_label).float()
test_accuracy = torch.mean(correct_mask).cpu()
model.test_accuracy += test_accuracy * input_label.size()[0]
model.test_loss /= batch_amount
model.test_accuracy /= batch_amount
if config.test:
predictions = [x.item() for x in predictions]
labels = [x.item() for x in labels]
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
else:
if model.test_accuracy.item() > best_accuracy:
best_accuracy = model.test_accuracy.item()
loss = model.test_loss.item()
acc = model.test_accuracy.item()
log(config.log_file,
'Tested network. So far best: {}'.format(best_accuracy))
log(config.log_file,
"TESTING EPOCH {} acc: {} loss: {}".format(epoch, acc, loss))
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
return best_accuracy
def test(config):
log_dir = os.path.join(config.log_dir, config.name + '_stage_2')
val_path = os.path.join(config.data, "*/test")
val_dataset = MultiviewImgDataset(val_path,
scale_aug=False,
rot_aug=False,
num_views=config.num_views)
val_loader = torch.utils.data.DataLoader(
val_dataset,
batch_size=config.stage2_batch_size,
shuffle=False,
num_workers=0)
pretraining = not config.no_pretraining
cnet = SVCNN(config.name,
nclasses=config.num_classes,
cnn_name=config.cnn_name,
pretraining=pretraining)
cnet_2 = MVCNN(config.name,
cnet,
nclasses=config.num_classes,
cnn_name=config.cnn_name,
num_views=config.num_views)
cnet_2.load(
os.path.join(log_dir, config.snapshot_prefix + str(config.weights)))
optimizer = optim.Adam(cnet_2.parameters(),
lr=config.learning_rate,
weight_decay=config.weight_decay,
betas=(0.9, 0.999))
trainer = ModelNetTrainer(cnet_2,
None,
val_loader,
optimizer,
nn.CrossEntropyLoss(),
config,
log_dir,
num_views=config.num_views)
labels, predictions = trainer.update_validation_accuracy(config.weights,
test=True)
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def eval(config, solver, epoch=0):
acc = 0
loss = 0
labels = []
predictions = []
test_count = get_dataset_size(config, 'test')
keys = solver.test_nets[0].blobs.keys()
batch_size = (
solver.test_nets[0].blobs['label_data_1_split_0'].data.shape[0])
test_iters = test_count / batch_size
for i in range(test_iters):
solver.test_nets[0].forward()
acc += solver.test_nets[0].blobs['accuracy'].data
loss += solver.test_nets[0].blobs['loss'].data
probs = solver.test_nets[0].blobs['ip2'].data
predictions += list(np.argmax(np.array(probs), axis=1))
labels += list(solver.test_nets[0].blobs['label'].data)
solver.test_nets[0].forward()
acc += solver.test_nets[0].blobs['accuracy'].data
loss += solver.test_nets[0].blobs['loss'].data
probs = solver.test_nets[0].blobs['ip2'].data
predictions += list(np.argmax(np.array(probs),
axis=1))[0:test_count % batch_size]
labels += list(solver.test_nets[0].blobs['label'].data)[0:test_count %
batch_size]
acc /= test_iters + 1
loss /= test_iters + 1
if not config.test:
print("Accuracy: {:.3f}".format(acc))
print("Loss: {:.3f}".format(loss))
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
else:
print("----------------------")
import Evaluation_tools as et
labels = [int(l) for l in labels]
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def evaluate(x_test, y_test, cfg, tfuncs, tvars, config, epoch=0):
log(config.log_file, "testing")
n_rotations = cfg['n_rotations']
chunk_size = n_rotations * cfg['batches_per_chunk']
num_chunks = int(math.ceil(float(len(x_test)) / chunk_size))
labels = []
test_class_error = []
pred_array = []
losses = []
accs = []
for chunk_index in xrange(num_chunks):
upper_range = min(len(y_test), (chunk_index + 1) * chunk_size) #
x_shared = np.asarray(x_test[chunk_index *
chunk_size:upper_range, :, :, :, :],
dtype=np.float32)
y_shared = np.asarray(y_test[chunk_index * chunk_size:upper_range],
dtype=np.float32)
num_batches = int(math.ceil(float(len(x_shared)) / n_rotations))
tvars['X_shared_'].set_value(4.0 * x_shared - 1.0, borrow=True)
tvars['y_shared_'].set_value(y_shared, borrow=True)
for bi in xrange(num_batches):
[batch_loss, batch_test_class_error, pred, raw_pred,
y] = tfuncs['test_function'](bi)
losses.append(batch_loss)
accs.append(batch_test_class_error)
pred_array.append(np.array(pred))
labels.append(y[0])
loss, acc = [float(np.mean(losses)), 1.0 - float(np.mean(accs))]
predictions = list(pred_array)
log(
config.log_file,
'EVALUTAION: epoch: {0:^3d}, loss: {2:.6f}, acc: {3:.5f}'.format(
epoch, loss, acc))
if not config.test:
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
else:
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def test(dataset, config):
print('test() called')
weights = config.weights
V = config.num_views
batch_size = config.batch_size
ckptfile = os.path.join(config.log_dir,
config.snapshot_prefix + str(weights))
data_size = dataset.size()
print('dataset size:', data_size)
with tf.Graph().as_default():
global_step = tf.Variable(0, trainable=False)
view_ = tf.placeholder('float32',
shape=(None, V, 227, 227, 3),
name='im0')
y_ = tf.placeholder('int64', shape=(None), name='y')
keep_prob_ = tf.placeholder('float32')
fc8 = model.inference_multiview(view_, config.num_classes, keep_prob_)
loss = model.loss(fc8, y_)
#train_op = model.train(loss, global_step, data_size)
prediction = model.classify(fc8)
placeholders = [view_, y_, keep_prob_, prediction, loss]
saver = tf.train.Saver(tf.all_variables())
init_op = tf.global_variables_initializer()
sess = tf.Session(config=tf.ConfigProto(log_device_placement=False))
saver.restore(sess, ckptfile)
print('restore variables done')
print("Start testing")
print("Size:", data_size)
print("It'll take", int(math.ceil(data_size / batch_size)),
"iterations.")
acc, _, predictions, labels = _test(dataset, config, sess,
placeholders)
print('acc:', acc * 100)
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def _evaluate(net, sess, test_data, config, epoch=0):
labels = []
predictions = []
losses = []
acc = 0
count = 0
while True:
batch_images, batch_labels, reset = test_data.next_batch(
config.batch_size)
if reset:
break
logits, loss = net.test(batch_images, batch_labels, sess)
losses.append(loss)
for i in range(len(batch_labels) / config.num_views):
endindex = i * config.num_views + config.num_views
prediction = np.argmax(
np.sum(logits[i * config.num_views:endindex], axis=0))
predictions.append(prediction)
label = batch_labels[i * config.num_views]
labels.append(label)
if label == prediction:
acc += 1
else:
acc += 0
count += 1
acc = acc / float(count)
log(config.log_file,
"EVALUATING epoch {} - acc: {} loss: {}".format(epoch, acc, loss))
if not config.test:
loss = np.mean(losses)
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
else:
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
def evaluate_ensemble(all_preds, labels):
all_preds = np.array(all_preds)
summed_preds = np.sum(all_preds, axis=0)
final_predictions = np.argmax(summed_preds, 1)
return final_predictions
### TODO: Clean this up and add the necessary arguments to enable all of the options we want.
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('model',
help='path to file containing model definition')
parser.add_argument('data',
default="/data/converted",
help='path to data folder')
parser.add_argument('--log_dir',
default="logs",
help='path to data folder')
parser.add_argument('--weights', type=int, help='number of model to test')
args = parser.parse_args()
file = args.model
predictions, labels = test(args)
import Evaluation_tools as et
eval_file = os.path.join(args.log_dir, 'vrnens.txt')
et.write_eval_file(args.data, eval_file, predictions, labels, 'VRNENS')
et.make_matrix(args.data, eval_file, args.log_dir)
def main(argv):
pycaffe_dir = caffe_root + 'python/'
parser = argparse.ArgumentParser()
# Required arguments: input and output files.
parser.add_argument("--input_file",
default="/data/converted/testrotnet.txt",
help="text file containg the image paths")
# Optional arguments.
parser.add_argument(
"--model_def",
default="./Training/rotationnet_modelnet40_case1_solver.prototxt",
help="Model definition file.")
parser.add_argument('--weights', type=int, default=-1)
parser.add_argument('--views', type=int, default=12)
parser.add_argument('--log_dir', default='logs', type=str)
parser.add_argument(
"--center_only",
action='store_true',
default=False,
help="Switch for prediction from center crop alone instead of " +
"averaging predictions across crops (default).")
parser.add_argument(
"--images_dim",
default='227,227',
help="Canonical 'height,width' dimensions of input images.")
parser.add_argument(
"--mean_file",
default=os.path.join(caffe_root,
'data/ilsvrc12/imagenet_mean.binaryproto'),
help="Data set image mean of H x W x K dimensions (np array). " +
"Set to '' for no mean subtraction.")
parser.add_argument(
"--input_scale",
type=float,
default=255,
help="Multiply input features by this scale before input to net")
parser.add_argument(
"--channel_swap",
default='2,1,0',
help="Order to permute input channels. The default converts " +
"RGB -> BGR since BGR is the Caffe default by way of OpenCV.")
args = parser.parse_args()
args.pretrained_model = os.path.join(
args.log_dir, 'case1_iter_' + str(args.weights) + '.caffemodel')
image_dims = [int(s) for s in args.images_dim.split(',')]
channel_swap = [int(s) for s in args.channel_swap.split(',')]
if args.mean_file:
mean = get_mean(args.mean_file)
caffe.set_mode_gpu()
# Make classifier.
classifier = caffe.Classifier(args.model_def,
args.pretrained_model,
image_dims=image_dims,
mean=mean,
input_scale=1.0,
raw_scale=255.0,
channel_swap=channel_swap)
listfiles, labels = read_lists(args.input_file)
#dataset = Dataset(listfiles, labels, subtract_mean=False, V=20)
# Load image file.
args.input_file = os.path.expanduser(args.input_file)
preds = []
labels = [int(label) for label in labels]
total = len(listfiles)
views = args.views
batch = 8 * views
for i in range(len(listfiles) / (batch * views)):
#im_files = [line.rstrip('\n') for line in open(listfiles[views*i+j])]
im_files = listfiles[i * batch * views:(i + 1) * batch * views]
#labels.append(int(im_files[0]))
#im_files = im_files[2:]
inputs = [caffe.io.load_image(im_f) for im_f in im_files]
predictions = classifier.predict(inputs, not args.center_only)
classified = classify(predictions)
preds.append(classified)
print(classified)
import Evaluation_tools as et
data = '/data'
logs = '/logs'
eval_file = os.path.join(logs, 'rotnet.txt')
et.write_eval_file(data, eval_file, preds, labels, 'ROTNET')
et.make_matrix(data, eval_file, logs)
def eval_one_epoch(config, sess, ops, epoch=0):
is_training = False
num_votes = config.num_votes
total_seen = 0
loss_sum = 0
predictions = []
labels = []
all = 0
for fn in range(len(TEST_FILES)):
log_string('----' + str(fn) + '-----')
current_data, current_label = provider.loadDataFile(TEST_FILES[fn])
current_data = current_data[:, 0:config.num_points, :]
current_label = np.squeeze(current_label)
file_size = current_data.shape[0]
num_batches = file_size // config.batch_size + 1
for batch_idx in range(num_batches):
start_idx = batch_idx * config.batch_size
end_idx = (batch_idx + 1) * config.batch_size
cur_batch_size = min(end_idx - start_idx,
config.batch_size - end_idx + file_size)
if cur_batch_size < config.batch_size:
placeholder_data = np.zeros(
([config.batch_size] + (list(current_data.shape))[1:]))
placeholder_data[0:cur_batch_size, :, :] = current_data[
start_idx:end_idx, :, :]
placeholder_labels = np.zeros((config.batch_size))
placeholder_labels[0:cur_batch_size] = current_label[
start_idx:end_idx]
batch_labels = placeholder_labels
batch_data = placeholder_data
else:
batch_data = current_data[start_idx:end_idx, :, :]
batch_labels = current_label[start_idx:end_idx]
# Aggregating BEG
batch_loss_sum = 0 # sum of losses for the batch
batch_pred_sum = np.zeros(
(config.batch_size, config.num_classes)) # score for classes
for vote_idx in range(num_votes):
rotated_data = provider.rotate_point_cloud_by_angle(
batch_data, vote_idx / float(num_votes) * np.pi * 2)
feed_dict = {
ops['pointclouds_pl']: rotated_data,
ops['labels_pl']: batch_labels,
ops['is_training_pl']: is_training
}
loss_val, pred_val = sess.run([ops['loss'], ops['pred']],
feed_dict=feed_dict)
batch_pred_sum += pred_val
batch_loss_sum += (loss_val * cur_batch_size /
float(num_votes))
pred_val = np.argmax(batch_pred_sum, 1)
predictions += pred_val.tolist()[0:cur_batch_size]
labels += current_label[start_idx:end_idx].tolist()
total_seen += cur_batch_size
loss_sum += batch_loss_sum
loss = loss_sum / float(total_seen)
acc = sum([
1 if predictions[i] == labels[i] else 0
for i in range(len(predictions))
]) / float(len(predictions))
print(loss)
print(acc)
if config.test:
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
else:
log_string('eval mean loss: %f' % loss)
LOSS_LOGGER.log(loss, epoch, "eval_loss")
log_string('eval accuracy: %f' % acc)
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
def eval_one_epoch(config, sess, ops, topk=1, epoch=0):
is_training = False
# Make sure batch data is of same size
cur_batch_data = np.zeros(
(config.batch_size, config.num_points, TEST_DATASET.num_channel()))
cur_batch_label = np.zeros((config.batch_size), dtype=np.int32)
total_correct = 0
total_seen = 0
loss_sum = 0
batch_idx = 0
shape_ious = []
predictions = []
labels = []
while TEST_DATASET.has_next_batch():
batch_data, batch_label = TEST_DATASET.next_batch(augment=False)
bsize = batch_data.shape[0]
# for the last batch in the epoch, the bsize:end are from last batch
cur_batch_data[0:bsize, ...] = batch_data
cur_batch_label[0:bsize] = batch_label
batch_pred_sum = np.zeros(
(config.batch_size, config.num_classes)) # score for classes
for vote_idx in range(config.num_votes):
# Shuffle point order to achieve different farthest samplings
shuffled_indices = np.arange(config.num_points)
np.random.shuffle(shuffled_indices)
if config.normal:
rotated_data = provider.rotate_point_cloud_by_angle_with_normal(
cur_batch_data[:, shuffled_indices, :],
vote_idx / float(config.num_votes) * np.pi * 2)
else:
rotated_data = provider.rotate_point_cloud_by_angle(
cur_batch_data[:, shuffled_indices, :],
vote_idx / float(config.num_votes) * np.pi * 2)
feed_dict = {
ops['pointclouds_pl']: rotated_data,
ops['labels_pl']: cur_batch_label,
ops['is_training_pl']: is_training
}
loss_val, pred_val = sess.run([ops['loss'], ops['pred']],
feed_dict=feed_dict)
batch_pred_sum += pred_val
pred_val = np.argmax(batch_pred_sum, 1)
correct = np.sum(pred_val[0:bsize] == batch_label[0:bsize])
predictions += pred_val[0:bsize].tolist()
labels += batch_label[0:bsize].tolist()
total_correct += correct
total_seen += bsize
loss_sum += loss_val
batch_idx += 1
loss = (loss_sum / float(batch_idx))
acc = (total_correct / float(total_seen))
log(config.log_file,
"EVALUATING epoch {} - loss: {} acc: {} ".format(epoch, loss, acc))
if config.test:
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, labels,
config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
else:
LOSS_LOGGER.log(loss, epoch, "eval_loss")
ACC_LOGGER.log(acc, epoch, "eval_accuracy")
TEST_DATASET.reset()
return total_correct / float(total_seen)
net = KDNET(config)
if config.weights != -1:
weights = config.weights
load_weights(os.path.join(config.log_dir, config.snapshot_prefix+str(weights)), net.KDNet['output'])
print("Loaded weights")
if config.test:
print("Start testing")
_, predictions = acc_fun(net, test_vertices, test_faces, test_nFaces, test_labels, mode='test',config=config)
acc = 100.*(predictions == test_labels).sum()/len(test_labels)
print('Eval accuracy: {}'.format(acc))
import Evaluation_tools as et
eval_file = os.path.join(config.log_dir, '{}.txt'.format(config.name))
et.write_eval_file(config.data, eval_file, predictions, test_labels, config.name)
et.make_matrix(config.data, eval_file, config.log_dir)
else:
print("Start training")
LOSS_LOGGER = Logger("{}_loss".format(config.name))
ACC_LOGGER = Logger("{}_acc".format(config.name))
start_epoch = 0
if config.weights != -1:
ld = config.log_dir
WEIGHTS = config.weights
ckptfile = os.path.join(ld,config.snapshot_prefix+str(WEIGHTS))
start_epoch = WEIGHTS + 1
ACC_LOGGER.load((os.path.join(ld,"{}_acc_train_accuracy.csv".format(config.name)),os.path.join(ld,"{}_acc_eval_accuracy.csv".format(config.name))), epoch = WEIGHTS)
LOSS_LOGGER.load((os.path.join(ld,"{}_loss_train_loss.csv".format(config.name)), os.path.join(ld,'{}_loss_eval_loss.csv'.format(config.name))), epoch = WEIGHTS)
begin = start_epoch
