def test(opt):
opt.nThreads = 1 # test code only supports nThreads = 1
opt.batchSize = 1 # test code only supports batchSize = 1
opt.serial_batches = True # no shuffle
opt.no_flip = True # no flip
data_loader = CreateDataLoader(opt)
dataset = data_loader.load_data()
dataset_size = len(data_loader)
print('#test batches = %d' % (int(dataset_size / len(opt.sort_order))))
visualizer = Visualizer(opt)
model = create_model(opt)
model.eval()
# create webpage
if opt.random_seed != -1:
exp_dir = '%s_%s_seed%s' % (opt.phase, opt.which_epoch, str(opt.random_seed))
else:
exp_dir = '%s_%s' % (opt.phase, opt.which_epoch)
web_dir = os.path.join(opt.results_dir, opt.name, exp_dir)
if opt.traverse or opt.deploy:
if opt.traverse:
out_dirname = 'traversal'
else:
out_dirname = 'deploy'
output_dir = os.path.join(web_dir,out_dirname)
if not os.path.isdir(output_dir):
os.makedirs(output_dir)
for image_path in opt.image_path_list:
print(image_path)
data = dataset.dataset.get_item_from_path(image_path)
visuals = model.inference(data)
if opt.traverse and opt.make_video:
out_path = os.path.join(output_dir, os.path.splitext(os.path.basename(image_path))[0] + '.mp4')
visualizer.make_video(visuals, out_path)
elif opt.traverse or (opt.deploy and opt.full_progression):
if opt.traverse and opt.compare_to_trained_outputs:
out_path = os.path.join(output_dir, os.path.splitext(os.path.basename(image_path))[0] + '_compare_to_{}_jump_{}.png'.format(opt.compare_to_trained_class, opt.trained_class_jump))
else:
out_path = os.path.join(output_dir, os.path.splitext(os.path.basename(image_path))[0] + '.png')
visualizer.save_row_image(visuals, out_path, traverse=opt.traverse)
else:
out_path = os.path.join(output_dir, os.path.basename(image_path[:-4]))
visualizer.save_images_deploy(visuals, out_path)
else:
webpage = html.HTML(web_dir, 'Experiment = %s, Phase = %s, Epoch = %s' % (opt.name, opt.phase, opt.which_epoch))
# test
for i, data in enumerate(dataset):
if i >= opt.how_many:
break
visuals = model.inference(data)
img_path = data['Paths']
rem_ind = []
for i, path in enumerate(img_path):
if path != '':
print('process image... %s' % path)
else:
rem_ind += [i]
for ind in reversed(rem_ind):
del img_path[ind]
visualizer.save_images(webpage, visuals, img_path)
webpage.save()
# upload your image (the code supports only a single image at a time)
from google.colab import files
uploaded = files.upload()
for filename in uploaded.keys():
img_path = filename
print('User uploaded file "{name}"'.format(name=filename))
"""Finally, we preprocess the image, run the network, and save the result."""
data = dataset.dataset.get_item_from_path(img_path)
visuals = model.inference(data)
os.makedirs('results', exist_ok=True)
out_path = os.path.join('results', os.path.splitext(img_path)[0].replace(' ', '_') + '.mp4')
visualizer.make_video(visuals, out_path)
"""Let's display at the results.
NOTE: if you're using chrome, uncomment the lines below. For some reason, mp4 files won't display on chrome browser, so we need to convert to webm.
"""
use_webm = False
# For some unknown reason the mp4 video is not displayed on chrome
# If you have chrome, uncomment the following lines to convert the
# result to webm for display purposes
!pip3 install webm
webm_out_path = os.path.join('results', os.path.splitext(img_path)[0].replace(' ', '_') + '.webm')
!webm -i $out_path $webm_out_path
use_webm = True
