def process(plt, dir, depth, rgb):
#extract depthmap
with zipfile.ZipFile(dir + '/depth/' + depth, 'r') as zip_ref:
zip_ref.extractall('.')
utils.parseData('data')
#read rgb data
global im_array
if rgb:
width = utils.getWidth()
height = utils.getHeight()
pil_im = Image.open(dir + '/rgb/' + rgb)
pil_im = pil_im.resize((width, height), Image.ANTIALIAS)
im_array = np.asarray(pil_im)
else:
im_array = 0
#parse calibration
global calibration
calibration = utils.parseCalibration(dir + '/camera_calibration.txt')
def main():
df = parseData(sys.argv[1])
#print(df)
df = processData(df)
#print(df)
drawFig(df, 'time')
platform = 'android' # Keyword (separated by either blanks or by the following logic commands: OR/AND/NOT keyword = 'diabetes OR mellitus' # App categories (separated by commas) categories = 'MEDICAL, LIFESTYLE, EDUCATION, HEALTH_AND_FITNESS' # Token of 42Matters token = '6dc3eba38263b374e06986f69c876c3ea6cb2f9f' # App's metadata language (all languages if argument is in blank) langs = 'en' con = connection.Connection() con.set_categories(categories) con.set_keyword(keyword) con.set_langs(langs) con.set_os(platform) con.set_token(token) data = con.get_data() utils = utils.Utils() result = utils.parseData(data) csv_file_name = 'result.csv' csv_export = csvExport.csvExport() csv_export.set_file_name(csv_file_name) csv_export.write_to_csv(result)
args.init_lr,
betas=(args.beta_1, args.beta_2))
scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer,
milestones=args.milestones,
gamma=args.lr_decay,
last_epoch=-1)
if args.masked_loss:
print('using mask')
criterion = utils.Criterion_mask(args)
else:
criterion = utils.Criterion(args)
model.train()
for epoch in range(1, args.epochs + 1):
print('---- Start Training Epoch %d: %d batches ----' %
(epoch, len(train_loader)))
scheduler.step()
for i, sample in enumerate(train_loader):
data = utils.parseData(args, sample, 'train')
input = [data['input']]
if args.in_light:
input.append(data['l'])
output = model(input)
optimizer.zero_grad()
loss = criterion.forward(output, data['tar'])
criterion.backward()
optimizer.step()
print("Loss in epoch %d: %.3f" % (epoch, loss))
torch.save(model, './TrainedModels/model_new.pth.tar')
print("saved the model")
test_set = DiLiGenT_main(args, 'test')
test_loader = torch.utils.data.DataLoader(test_set,
batch_size=args.test_batch,
num_workers=args.workers,
pin_memory=args.cuda,
shuffle=False)
model.eval()
print('---- Testing for %d images - DiLiGent Dataset ----' %
(len(test_loader)))
err_mean = 0
with torch.no_grad():
for i, sample in enumerate(test_loader):
data = utils.parseData(args, sample, 'test')
input = [data['input']]
if args.in_light:
input.append(data['l'])
output = model(input)
acc = utils.errorPred(data['tar'].data, output.data, data['m'].data)
err_mean = err_mean + acc
print('error: %.3f' % (acc))
result = (output.data + 1) / 2
result_masked = result * data['m'].data.expand_as(output.data)
save_path = './Results/' + 'img8_mask_%d.png' % (i + 1)
tv.utils.save_image(result_masked, save_path)
print('saved image %d' % (i + 1))
print('------------ mean error: %.3f ------------' %
#import src.api.utils as utils
import config
import utils
app = Flask(__name__)
logger = utils.setup_logger()
@app.route('/')
def hello_world():
return 'Running Correctly!'
if __name__ == '__main__':
if config.STARTUP["DOWNLOAD"]:
utils.downloadAllData()
if config.STARTUP["EXTRACT"]:
utils.extractData()
if config.STARTUP["PARSE"]:
data = utils.parseData()
if config.STARTUP["REBUILD_DB"]:
utils.buildDB(data)
#app.run(debug=config.DEBUG, host = config.HOST)
app.run(host=config.HOST)