def handle_data():
global frameCount
global filePath
global fileName
if request.method == 'POST':
if request.form.get('pcaselect'):
use_eigenface = True
use_lbp = False
model = "none"
loss = "none"
pcaselect = "yes"
lbpselect = "no"
elif request.form.get('lbpselect'):
use_eigenface = False
use_lbp = True
model = "none"
loss = "none"
pcaselect = "no"
lbpselect = "yes"
else:
use_eigenface = False
use_lbp = False
pcaselect = "no"
lbpselect = "no"
model = request.form['model']
loss = request.form['loss']
model_pth = 'backbone/{}_{}.pth'.format(model, loss)
if model == '10':
args.backbone = 'spherenet10'
elif model == '20':
args.backbone = 'spherenet20'
elif model == '64':
args.backbone = 'spherenet64'
netModel = CreateModel(args)
netModel.backbone.load_state_dict(torch.load(model_pth))
if 'file' not in request.files:
if frameCount > 1:
image = Image.open(filePath)
landmarks = get_landmarks(image)
aligned_face = get_alignedface(image, landmarks)
aligned_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'aligned_face.jpg')
cv2.imwrite(aligned_face_path, aligned_face)
if use_eigenface:
recon_face, eigenface_result = eigenface_evaluation(
aligned_face)
recon_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'reconstructed_face.jpg')
cv2.imwrite(recon_face_path, recon_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = eigenface_result
elif use_lbp:
lbp_face = local_binary_pattern(
cv2.cvtColor(aligned_face, cv2.COLOR_BGR2GRAY), 8, 1)
lbp_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'lbp_face.jpg')
cv2.imwrite(lbp_face_path, lbp_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = lbp_evaluation(
aligned_face)
else:
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = evaluation(
netModel.backbone, aligned_face)
face_1_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_1.jpg')
face_2_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_2.jpg')
face_3_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_3.jpg')
cv2.imwrite(face_1_path, face_1)
cv2.imwrite(face_2_path, face_2)
cv2.imwrite(face_3_path, face_3)
imageName = fileName.rsplit('_', 1)[0]
name1 = name1.rsplit('_', 1)[0]
name2 = name2.rsplit('_', 1)[0]
name3 = name3.rsplit('_', 1)[0]
if use_eigenface:
mean_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'meanface.jpg')
eigenface1 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface1.jpg')
eigenface2 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface2.jpg')
eigenface3 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface3.jpg')
eigenface4 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface4.jpg')
eigenface5 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface5.jpg')
eigenface6 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface6.jpg')
eigenface7 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface7.jpg')
eigenface8 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface8.jpg')
eigenface9 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface9.jpg')
return render_template('pca_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
mean_face=mean_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
eigenface1=eigenface1,
eigenface2=eigenface2,
eigenface3=eigenface3,
eigenface4=eigenface4,
eigenface5=eigenface5,
eigenface6=eigenface6,
eigenface7=eigenface7,
eigenface8=eigenface8,
eigenface9=eigenface9,
recon_face=recon_face_path,
num_faces=args.num_faces)
elif use_lbp:
return render_template('lbp_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
lbp_face=lbp_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
else:
return render_template('net_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
else:
flash('No file part')
return redirect(request.url)
file = request.files['file']
if file.filename == '':
if frameCount > 1:
image = Image.open(filePath)
landmarks = get_landmarks(image)
aligned_face = get_alignedface(image, landmarks)
aligned_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'aligned_face.jpg')
cv2.imwrite(aligned_face_path, aligned_face)
if use_eigenface:
recon_face, eigenface_result = eigenface_evaluation(
aligned_face)
recon_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'reconstructed_face.jpg')
cv2.imwrite(recon_face_path, recon_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = eigenface_result
elif use_lbp:
lbp_face = local_binary_pattern(
cv2.cvtColor(aligned_face, cv2.COLOR_BGR2GRAY), 8, 1)
lbp_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'lbp_face.jpg')
cv2.imwrite(lbp_face_path, lbp_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = lbp_evaluation(
aligned_face)
else:
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = evaluation(
netModel.backbone, aligned_face)
face_1_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_1.jpg')
face_2_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_2.jpg')
face_3_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_3.jpg')
cv2.imwrite(face_1_path, face_1)
cv2.imwrite(face_2_path, face_2)
cv2.imwrite(face_3_path, face_3)
imageName = fileName.rsplit('_', 1)[0]
name1 = name1.rsplit('_', 1)[0]
name2 = name2.rsplit('_', 1)[0]
name3 = name3.rsplit('_', 1)[0]
# return '', 204
# return render_template('gui.html', accuracy='2', image=file_path)
if use_eigenface:
mean_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'meanface.jpg')
eigenface1 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface1.jpg')
eigenface2 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface2.jpg')
eigenface3 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface3.jpg')
eigenface4 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface4.jpg')
eigenface5 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface5.jpg')
eigenface6 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface6.jpg')
eigenface7 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface7.jpg')
eigenface8 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface8.jpg')
eigenface9 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface9.jpg')
return render_template('pca_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
mean_face=mean_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
eigenface1=eigenface1,
eigenface2=eigenface2,
eigenface3=eigenface3,
eigenface4=eigenface4,
eigenface5=eigenface5,
eigenface6=eigenface6,
eigenface7=eigenface7,
eigenface8=eigenface8,
eigenface9=eigenface9,
recon_face=recon_face_path,
num_faces=args.num_faces)
elif use_lbp:
return render_template('lbp_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
lbp_face=lbp_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
else:
return render_template('net_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
# return '', 204
else:
flash('No selected file')
return redirect(request.url)
if file and allowed_file(file.filename):
frameCount = frameCount + 1
file_name = secure_filename(file.filename)
file_path = os.path.join(app.config['UPLOAD_FOLDER'], file_name)
file.save(file_path)
filePath = file_path
fileName = file_name
image = Image.open(file_path)
landmarks = get_landmarks(image)
aligned_face = get_alignedface(image, landmarks)
aligned_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'aligned_face.jpg')
cv2.imwrite(aligned_face_path, aligned_face)
if use_eigenface:
recon_face, eigenface_result = eigenface_evaluation(
aligned_face)
recon_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'reconstructed_face.jpg')
cv2.imwrite(recon_face_path, recon_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = eigenface_result
elif use_lbp:
lbp_face = local_binary_pattern(
cv2.cvtColor(aligned_face, cv2.COLOR_BGR2GRAY), 8, 1)
lbp_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'lbp_face.jpg')
cv2.imwrite(lbp_face_path, lbp_face)
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = lbp_evaluation(
aligned_face)
else:
face_1, face_2, face_3, name1, name2, name3, score1, score2, score3 = evaluation(
netModel.backbone, aligned_face)
face_1_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_1.jpg')
face_2_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_2.jpg')
face_3_path = os.path.join(app.config['UPLOAD_FOLDER'],
'face_3.jpg')
cv2.imwrite(face_1_path, face_1)
cv2.imwrite(face_2_path, face_2)
cv2.imwrite(face_3_path, face_3)
imageName = file_name.rsplit('_', 1)[0]
name1 = name1.rsplit('_', 1)[0]
name2 = name2.rsplit('_', 1)[0]
name3 = name3.rsplit('_', 1)[0]
# return '', 204
# return render_template('gui.html', accuracy='2', image=file_path)
if use_eigenface:
mean_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'meanface.jpg')
eigenface1 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface1.jpg')
eigenface2 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface2.jpg')
eigenface3 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface3.jpg')
eigenface4 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface4.jpg')
eigenface5 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface5.jpg')
eigenface6 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface6.jpg')
eigenface7 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface7.jpg')
eigenface8 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface8.jpg')
eigenface9 = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface9.jpg')
return render_template('pca_result.html',
imageName=imageName,
input_face=filePath,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
mean_face=mean_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
eigenface1=eigenface1,
eigenface2=eigenface2,
eigenface3=eigenface3,
eigenface4=eigenface4,
eigenface5=eigenface5,
eigenface6=eigenface6,
eigenface7=eigenface7,
eigenface8=eigenface8,
eigenface9=eigenface9,
recon_face=recon_face_path,
num_faces=args.num_faces)
elif use_lbp:
return render_template('lbp_result.html',
imageName=imageName,
input_face=file_path,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
lbp_face=lbp_face_path,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
else:
return render_template('net_result.html',
imageName=imageName,
input_face=file_path,
aligned_face=aligned_face_path,
image1=face_1_path,
name1=name1,
score1=score1,
image2=face_2_path,
name2=name2,
score2=score2,
image3=face_3_path,
name3=name3,
score3=score3,
pcaselect=pcaselect,
lbpselect=lbpselect,
model=model,
loss=loss,
num_faces=args.num_faces)
else:
return redirect(request.url)
eval_start_time = time.time()
for step, batch in enumerate(batches):
if step >= steps_per_epoch:
break
labels.extend(batch['label'])
probs.extend(model.forward(batch['X']))
util.over_print("Step: %d/%d \tTime: %f\r" % (step+1, steps_per_epoch, time.time()-eval_start_time))
loss, roc_auc, accuracy = util.measure_performance(labels, probs)
return loss, roc_auc, accuracy
with tf.Session(config=gpu_config) as sess:
dataloader = CreateDataLoader(opt, sess)
coord = tf.train.Coordinator()
dataloader.start_enqueue_threads_for_training(coord)
model = CreateModel(opt, sess, pos_weight=dataloader.pos_ratio)
logger = SimpleLogger(opt)
total_steps = -1
best_valid_roc_auc = 0
best_test_roc_auc = [0]*len(dataloader.test_datasets)
if opt.load_epoch == "best_test":
load_epoch = 0
else:
load_epoch = int(opt.load_epoch[-1])
for epoch in range(0, opt.nepoch):
# Skip epoch to continue training from specifed load epoch
if opt.continue_train and epoch < load_epoch:
def camresult():
global frameCount
if request.method == 'POST':
if request.form.get('pcaselect'):
use_eigenface = True
use_lbp = False
model = "none"
loss = "none"
pcaselect = "yes"
lbpselect = "no"
elif request.form.get('lbpselect'):
use_eigenface = False
use_lbp = True
model = "none"
loss = "none"
pcaselect = "no"
lbpselect = "yes"
else:
use_eigenface = False
use_lbp = False
pcaselect = "no"
lbpselect = "no"
model = request.form['model']
loss = request.form['loss']
model_pth = 'backbone/{}_{}.pth'.format(model, loss)
if model == '10':
args.backbone = 'spherenet10'
elif model == '20':
args.backbone = 'spherenet20'
elif model == '64':
args.backbone = 'spherenet64'
netModel = CreateModel(args)
netModel.backbone.load_state_dict(torch.load(model_pth))
data_url = request.form['data']
content = data_url.split(';')[1]
image_encoded = content.split(',')[1]
body = base64.decodebytes(image_encoded.encode('utf-8'))
webcam_img_path = os.path.join(app.config['UPLOAD_FOLDER'],
'webcam_img.jpg')
with open(webcam_img_path, "wb") as fh:
fh.write(body)
image = Image.open(webcam_img_path)
landmarks = get_landmarks(image)
aligned_face = get_alignedface(image, landmarks)
webcam_alignedimg_path = os.path.join(app.config['UPLOAD_FOLDER'],
'webcam_alignedimg.jpg')
cv2.imwrite(webcam_alignedimg_path, aligned_face)
if use_eigenface:
recon_face, eigenface_result = eigenface_evaluation(aligned_face)
recon_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'reconstructed_face.jpg')
cv2.imwrite(recon_face_path, recon_face)
face_1, face_2, face_3, name_1, name_2, name_3, score_1, score_2, score_3 = eigenface_result
elif use_lbp:
lbp_face = local_binary_pattern(
cv2.cvtColor(aligned_face, cv2.COLOR_BGR2GRAY), 8, 1)
lbp_face_path = os.path.join(app.config['UPLOAD_FOLDER'],
'lbp_face.jpg')
cv2.imwrite(lbp_face_path, lbp_face)
face_1, face_2, face_3, name_1, name_2, name_3, score_1, score_2, score_3 = lbp_evaluation(
aligned_face)
else:
face_1, face_2, face_3, name_1, name_2, name_3, score_1, score_2, score_3 = evaluation(
netModel.backbone, aligned_face)
face_1_path = os.path.join(app.config['UPLOAD_FOLDER'], 'face_1.jpg')
face_2_path = os.path.join(app.config['UPLOAD_FOLDER'], 'face_2.jpg')
face_3_path = os.path.join(app.config['UPLOAD_FOLDER'], 'face_3.jpg')
cv2.imwrite(face_1_path, face_1)
cv2.imwrite(face_2_path, face_2)
cv2.imwrite(face_3_path, face_3)
name1 = name_1.rsplit('_', 1)[0]
name2 = name_2.rsplit('_', 1)[0]
name3 = name_3.rsplit('_', 1)[0]
session['use_eigenface'] = use_eigenface
session['use_lbp'] = use_lbp
session['aligned_face'] = webcam_alignedimg_path
session['image1'] = face_1_path
session['name1'] = name1
session['score1'] = score_1
session['image2'] = face_2_path
session['name2'] = name2
session['score2'] = score_2
session['image3'] = face_3_path
session['name3'] = name3
session['score3'] = score_3
session['pcaselect'] = pcaselect
session['lbpselect'] = lbpselect
session['model'] = model
session['loss'] = loss
if use_eigenface:
session['mean_face_path'] = os.path.join(
app.config['UPLOAD_FOLDER'], 'meanface.jpg')
session['eigenface1'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface1.jpg')
session['eigenface2'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface2.jpg')
session['eigenface3'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface3.jpg')
session['eigenface4'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface4.jpg')
session['eigenface5'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface5.jpg')
session['eigenface6'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface6.jpg')
session['eigenface7'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface7.jpg')
session['eigenface8'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface8.jpg')
session['eigenface9'] = os.path.join(app.config['UPLOAD_FOLDER'],
'eigenface9.jpg')
session['recon_face_path'] = recon_face_path
return redirect(url_for('result_cam'))
elif use_lbp:
session['lbp_face'] = lbp_face_path
return redirect(url_for('result_cam'))
else:
return redirect(url_for('result_cam'))
else:
return redirect(request.url)
np.random.seed(opt.seed)
random.seed(opt.seed)
tf.set_random_seed(opt.seed)
result_dict = {
"split_0_result": None,
"split_1_result": None,
"split_2_result": None,
"split_3_result": None,
"split_4_result": None,
}
with tf.Session(config=gpu_config) as sess:
model = CreateModel(opt, sess)
data_loader = CreateDataLoader(opt, sess)
coord = tf.train.Coordinator()
data_loader.start_enqueue_threads_for_testing(coord)
data_loader.init_test_generator()
# Measure Performance on Test Datasets
print("Measuring Test Loss and Accuracy...")
test_roc_aucs = []
for split, dataset in enumerate(data_loader.test_datasets):
if not opt.load_epoch:
model.load("best_test_{}".format(split))
print("Data root: ", dataset['dataroot'])
from models.models import CreateModel
from common.loader import get_loader
from common.util import save_log
from arguments import train_args
from tqdm import tqdm
import val_verification
import val_identification
if __name__ == '__main__':
args = train_args.get_args()
dataloader = get_loader(args, 'training', args.bs)
model = CreateModel(args, class_num=dataloader.num_class)
model.train_setup(
) # setup the optimizer, scheduler and initialization for training
# model.eval()
# import hiddenlayer as hl
# import torch
# input = torch.randn(1, 3, 64, 64)
# graph = hl.build_graph(model.backbone, input)
# graph = graph.build_dot()
# graph.render('/Users/aaron/Desktop/spherenet.png', view=True, format='png')
# from torchsummary import summary
# summary(model.backbone, (3, 96, 96)) # summary(your_model, input_size=(channels, H, W))
pbar = tqdm(range(1, args.epochs + 1), ncols=0)
best_acc = 0
best_auc = 0.50
for epoch in pbar:
model.train()
import data_label
opt = TestOptions().parse()
opt.batchSize = 128
gpu_config = tf.ConfigProto(device_count={'GPU': opt.gpu_count})
dst = "./visualization/"
util.mkdirs(dst)
################################################
# Acquire prediction results on all samples #
################################################
with tf.Session(config=gpu_config) as sess:
model = CreateModel(opt, sess)
data_loader = CreateDataLoader(opt, sess)
coord = tf.train.Coordinator()
data_loader.start_enqueue_threads_for_testing(coord)
data_loader.init_test_generator()
y_true = []
y_pred = []
step_count = 0.0
for step, batch in enumerate(data_loader.test_datasets[0]['test_batch']):
probs = model.forward(batch['X'])
y_true.extend(batch['label'])
y_pred.extend(probs)
step_count += 1
util.over_print("Step count: {} \r".format(str(step_count)))
import torch
import torch.nn as nn
from models.models import CreateModel
from arguments import test_args
import pandas as pd
import test_identification
import test_verification
_backbone_model = '/Users/aaron/Desktop/11785-Intro to Deep Learning/Homework Part2/Homework2/src/checkpoints/0223-1/19_net_backbone.pth'
_criterion_model = '/Users/aaron/Desktop/11785-Intro to Deep Learning/Homework Part2/Homework2/src/checkpoints/0223-1/19_net_criterion.pth'
if __name__ == '__main__':
args = test_args.get_args()
model = CreateModel(args, class_num=args.class_num)
model.backbone.load_state_dict(torch.load(_backbone_model))
model.criterion.load_state_dict(torch.load(_criterion_model))
if torch.cuda.is_available():
if len(args.gpu_ids) > 1:
model.backbone = nn.DataParallel(model.backbone)
test_identification_id, test_identification_label = test_identification.run(
model, args)
d = {'id': test_identification_id, 'Category': test_identification_label}
df = pd.DataFrame(data=d)
df.to_csv('hw2p2_identification_test.csv', header=True, index=False)
print(
'Identification testing is done, result is saved to hw2p2_identification_test.csv'
)
test_verification_trial, test_verification_score = test_verification.run(