train_options.py

import argparse

class arguments():
    def __init__(self):
        self.argparser = argparse.ArgumentParser()
        self.initialize()
    
    def initialize(self):
        self.argparser.add_argument('--frame_dir', type=str, default='frame', help='path to frames')
        self.argparser.add_argument('--img_save_dir', type=str, default='results', help='path to storage generated feature maps if needed')
        self.argparser.add_argument('--n_epoch', type=int, default=1000, help='number of epochs')
        self.argparser.add_argument('--n_threads', type=int, default=1, help='number of threads for dataloader')
        self.argparser.add_argument('--batch_size', type=int, default=1, help='just batch size')
        self.argparser.add_argument('--learning_rate', type=float, default=0.005, help='learning rate')
        self.argparser.add_argument('--is_shuffle', type=bool, default=False, help='Do shuffle during loading data or not')
        self.argparser.add_argument('--visualize', type=bool, default=True, help='storage the flow in image type')        
        self.argparser.add_argument('--data_size', help='input data size')
        self.argparser.add_argument('--zfactor', type=float, default=0.5, help='factor for building the image piramid')    
        self.argparser.add_argument('--max_nscale', type=int, default=1, help='maximum number of scales for image piramid')    
        self.argparser.add_argument('--tau', type=float, default=0.25, help='time step')    
        self.argparser.add_argument('--lbda', type=float, default=0.15, help='weight parameter for the data term')    
        self.argparser.add_argument('--theta', type=float, default=0.3, help='weight parameter for (u - v)^2')    
        self.argparser.add_argument('--n_warps', type=int, default=1, help='number of warpings per scale')
        self.argparser.add_argument('--n_iters', type=int, default=30, help='maximum number of iterations for optimization')
        self.argparser.add_argument('--demo', help="just demo with original weights", action="store_true")

    def parse(self):
        self.args = self.argparser.parse_args()
        return self.args