HYTOS

from app import app

if __name__ == '__main__':

    #app.run(debug=False)

        self.use_rotation = True

        self.data_path = root_path

        self.xml_path = root_path + '/xml/'

        self.img_path_list = sorted([item for sublist in [glob.glob(self.img_path + ext) for ext in ["*.jpg", "*.png"]] for item in sublist])#sorted(glob.glob(os.path.join(self.img_path, '*.jpg')))

import argparse

import torch.nn as nn

from torch.utils.data import DataLoader

from src.utils import *

from src.loss import YoloLoss

from src.yolo_net import Yolo

            loss.backward()

            optimizer.step()

                print("Epoch: {}/{}, Iteration: {}/{}, Lr: {}, Loss:{:.5f} (Coord:{:.5f} Conf:{:.5f} Cls:{:.5f})".format

                    (epoch + 1, opt.num_epoches, iter + 1, num_iter_per_epoch, optimizer.param_groups[0]['lr'], loss,

                    loss_coord,loss_conf,loss_cls))

            loss_conf_ls = []

            loss_cls_ls = []

            for te_iter, te_batch in enumerate(test_generator):

                num_sample = len(te_label)

                if torch.cuda.is_available():

                    te_image = te_image.cuda()

                best_epoch = epoch

                print("SAVE MODEL")

                # for debug for each loss

                #torch.save(model, os.path.join(opt.saved_path, name + "_whole_model_" + str(save_count) + "_" + "{:.5f}".format(best_loss) + ".pth"))

                # save

                torch.save(model.state_dict(), os.path.join(opt.saved_path, name + "_only_params.pth"))

                save_count += 1

                # for debug for each loss

                torch.save(model.state_dict(), os.path.join(opt.saved_path, name + "_only_params_" + str(save_count) + "_" + "{:.5f}".format(te_loss) + ".pth"))

            # Early stopping

            if epoch - best_epoch > opt.es_patience > 0: