main_fullpipeline.py

from __future__ import print_function
import argparse
import os
import sys
import random
import torch
import torchvision
import torch.nn as nn
import torch.nn.parallel
import torch.optim as optim
import torch.utils.data
from torch.autograd import Variable
import torch.nn.functional as F
import numpy as np
import time
import math
import scipy.io
from metrics import *
from torch.utils.tensorboard import SummaryWriter
from tqdm import tqdm
from dataset_loader import OmniDepthDataset
import cv2
import spherical as S360
import supervision as L
from util import *
#import weight_init
from sync_batchnorm import convert_model
from CasStereoNet.models import psmnet_spherical_up_down_inv_depth
from CasStereoNet.models.loss import stereo_psmnet_loss
from network_resnet_v2 import ResNet360
#from network_resnet_v2 import ResNet360
# network_rectnet import RectNet
import matplotlib.pyplot as plot
import scipy.io
import csv

parser = argparse.ArgumentParser(description='PanoDepth')
parser.add_argument('--maxdisp', type=int ,default=192,
                    help='maxium disparity')
parser.add_argument('--model', default='psmnet',
                    help='select model')
parser.add_argument('--input_dir', default='/media/rtx2/DATA/Student_teacher_depth/stanford2d3d',
                    help='input data directory')
parser.add_argument('--trainfile', default='train_stanford2d3d.txt',
                    help='train file name')
parser.add_argument('--testfile', default='test_stanford2d3d.txt',
                    help='validation file name')
parser.add_argument('--epochs', type=int, default=150,
                    help='number of epochs to train')
parser.add_argument('--start_decay', type=int, default=60,
                    help='number of epoch for lr to start decay')
parser.add_argument('--start_learn', type=int, default=100,
                    help='number of iterations for stereo network to start learn')
parser.add_argument('--batch', type=int, default=8,
                    help='number of batch to train')
parser.add_argument('--visualize_interval', type=int, default=40,
                    help='number of batch to train')
parser.add_argument('--baseline', type=float, default=0.24,
                    help='image pair baseline distance')
parser.add_argument('--interval', type=float, default=0.5,
                    help='second stage interval')
parser.add_argument('--nlabels', type=str, default="48,24", 
                    help='number of labels')
parser.add_argument('--checkpoint', default= None,
                    help='load checkpoint path')
parser.add_argument('--save_checkpoint', default='./checkpoints',
                    help='save checkpoint path')
parser.add_argument('--visualize_path', default='./visualize_stanford2d3d_ours_48_24',
                    help='save checkpoint path')                    
parser.add_argument('--tensorboard_path', default='./logs',
                    help='tensorboard path')
parser.add_argument('--no-cuda', action='store_true', default=False,
                    help='disables CUDA training')
parser.add_argument('--real', action='store_true', default=False,
                    help='adapt to real world images in both training and validation')
parser.add_argument('--seed', type=int, default=1, metavar='S',
                    help='random seed (default: 1)')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()

# Save Checkpoint -------------------------------------------------------------
if not os.path.isdir(os.path.join(args.visualize_path, args.save_checkpoint)):
    os.makedirs(os.path.join(args.visualize_path, args.save_checkpoint))
    
# tensorboard Path -----------------------
writer_path = os.path.join(args.visualize_path, args.tensorboard_path)
writer = SummaryWriter(writer_path)

#-----------------------------------------

# Random Seed -----------------------------
torch.manual_seed(args.seed)
if args.cuda:
    torch.cuda.manual_seed(args.seed)
#------------------------------------------
input_dir = args.input_dir # Dataset location
train_file_list = args.trainfile # File with list of training files
val_file_list = args.testfile # File with list of validation files

#------------------------------------
result_view_dir = args.visualize_path
if not os.path.exists(result_view_dir):
    os.makedirs(result_view_dir) 
     
#-------------------------------------------------------------------
baseline = 0.24
baseline_direction = 'vertical'

batch_size = args.batch
maxdisp = args.maxdisp
nlabels = [int(nd) for nd in args.nlabels.split(",") if nd]
visualize_interval = args.visualize_interval
interval = args.interval
lr1 = 2e-4
lr2 = 5e-4
#-------------------------------------------------------------------
#data loaders
train_dataset = OmniDepthDataset(
		root_path=input_dir, 
        rotate=True,
        flip=True,
        gamma=True,
		path_to_img_list=train_file_list)

train_dataloader = torch.utils.data.DataLoader(
	dataset=train_dataset,
	batch_size=batch_size,
	shuffle=True,
	num_workers=8,
    pin_memory=True,
	drop_last=True)

val_dataset = OmniDepthDataset(
		root_path=input_dir, 
		path_to_img_list=val_file_list)

val_dataloader = torch.utils.data.DataLoader(
	dataset=val_dataset,
	batch_size=4,
	shuffle=False,
	num_workers=8,
	drop_last=True)

#----------------------------------------------------------
#first network, coarse depth estimation
# option 1, resnet 360 
num_gpu = torch.cuda.device_count()
first_network = ResNet360()
#first_network = ResNet360(conv_type='standard', norm_type='batchnorm', activation='relu', aspp=True)
#weight_init.initialize_weights(first_network, init="xavier", pred_bias=float(5.0))
#first_network = RectNet()
first_network = convert_model(first_network)
# option 2, spherical unet
#view_syn_network = SphericalUnet()
first_network = nn.DataParallel(first_network)
state_dict = torch.load('visualize_coarse_pretrain/checkpoints/checkpoint_latest.tar')
first_network.load_state_dict(state_dict)
first_network.cuda()
#----------------------------------------------------------

# stereo matching network ----------------------------------------------
if args.model == 'psmnet':
    stereo_network = psmnet_spherical_up_down_inv_depth.PSMNet(nlabels, 
        [1,0.5], True, 5, cr_base_chs=[32,32,16])
    stereo_network = convert_model(stereo_network)
else:
    print('Model Not Implemented!!!')
#----------------------------------------------------------

#-----------------------------------------------------------------------------
stereo_network = nn.DataParallel(stereo_network)
stereo_network.cuda()
#-------------------------------------------------

# Load Checkpoint -------------------------------
start_epoch = 0
if args.checkpoint is not None:
    state_dict = torch.load(args.checkpoint)
    first_network.load_state_dict(state_dict['state_dict1'])
    stereo_network.load_state_dict(state_dict['state_dict2'])
    start_epoch = state_dict['epoch']

print('## Batch size: {}'.format(batch_size))  
print('## learning rate 1: {}, learning rate 2: {}'.format(lr1, lr2))  
print('## Number of first model parameters: {}'.format(sum([p.data.nelement() for p in first_network.parameters()])))
print('## Number of stereo matching model parameters: {}'.format(sum([p.data.nelement() for p in stereo_network.parameters()])))
#--------------------------------------------------

# Optimizer ----------
#optimizer = optim.Adam(list(model.parameters())+list(view_syn_network.parameters()), 
#        lr=0.001, betas=(0.9, 0.999))
optimizer1 = optim.Adam(list(first_network.parameters()), 
        lr=lr1, betas=(0.9, 0.999))
optimizer2 = optim.Adam(list(stereo_network.parameters()), 
        lr=lr2, betas=(0.9, 0.999))
scheduler1 = optim.lr_scheduler.MultiStepLR(optimizer1, [20, 40, 60], 0.5)
scheduler2 = optim.lr_scheduler.MultiStepLR(optimizer2, [20, 40, 60], 0.5)
#---------------------

    
# Train Function -------------------
def train(target, render, depth_down, mask, batch_idx):
    #stereo_network.train()
    #mask = mask>0

    # mask value
    #mask = (disp_true < args.maxdisp) & (disp_true > 0)
    mask = mask>0
    mask.detach_()

    # Loss -------------------------------------------- 
    #outputs = stereo_network(render[0], target, baseline=baseline, direction=baseline_direction)
    outputs = stereo_network(render, target)
    #sgrid = S360.grid.create_spherical_grid(512).cuda()
    #disparity = S360.derivatives.dtheta_vertical(sgrid, depth_down.unsqueeze(1), baseline).squeeze(1)
    stereo_loss = stereo_psmnet_loss(outputs, depth_down, mask, dlossw=[0.5, 2.0])
    #loss = disp_loss
    #--------------------------------------------------
    output3 = outputs["stage2"]["pred"]
    #output3 = S360.derivatives.disparity_to_depth_vertical(sgrid, output3.unsqueeze(1), baseline).squeeze(1)
    gt = target[:,:3,:,:].detach().cpu().numpy()
    render_np = render.detach().cpu().numpy()
    depth = depth_down.detach().cpu().numpy()
    depth_prediction = output3.detach().cpu().numpy()
    depth_prediction[depth_prediction>8] = 0

    if batch_idx % visualize_interval == 0 and batch_idx > 0:
            gt_img = gt[0, :, :, :].transpose(1,2,0)
            depth_img = depth[0, :, :]
            #depth_down_img = cv2.normalize(depth_down_img,None,255.0,norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8UC1)
            depth_pred_img = depth_prediction[0, :, :]
            #depth_pred_img = cv2.normalize(depth_pred_img,None,255.0,norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8UC1)
            cv2.imwrite('{}/gt_{}.png'.format(result_view_dir, batch_idx), gt_img*255)
            render_img = render_np[0, :, :, :].transpose(1,2,0)
            cv2.imwrite('{}/render_{}.png'.format(result_view_dir, batch_idx), render_img*255)
            plot.imsave('{}/depth_gt_{}.png'.format(result_view_dir, batch_idx), depth_img, cmap="jet")
            plot.imsave('{}/depth_pred_{}.png'.format(result_view_dir, batch_idx), depth_pred_img, cmap="jet")

    return output3, stereo_loss

# Valid Function -----------------------
def val(target, render, depth, mask, batch_idx):
    stereo_network.eval() 
    sgrid = S360.grid.create_spherical_grid(512).cuda()
    with torch.no_grad():
        #outputs = stereo_network(render[0], target, baseline=baseline, direction=baseline_direction)
        outputs =stereo_network(render, target)
        output3 = outputs["stage2"]["pred"]
        #output3 = S360.derivatives.disparity_to_depth_vertical(sgrid, output3.unsqueeze(1), baseline).squeeze(1)

    gt = target[:,:3,:,:].detach().cpu().numpy()
    render_np = render.detach().cpu().numpy()

    depth = depth.detach().cpu().numpy()
    sgrid = S360.grid.create_spherical_grid(512).cuda()
    depth_prediction = output3.detach().cpu().numpy()
    depth_prediction[depth_prediction>8] = 0
    if batch_idx % 20 == 0 and batch_idx > 0:
            gt_img = gt[0, :, :, :].transpose(1,2,0)
            depth_img = depth[0, :, :]
            #depth_down_img = cv2.normalize(depth_down_img,None,255.0,norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8UC1)
            depth_pred_img = depth_prediction[0, :, :]
            #depth_pred_img = cv2.normalize(depth_pred_img,None,255.0,norm_type=cv2.NORM_MINMAX, dtype=cv2.CV_8UC1)
            cv2.imwrite('{}/gt_{}.png'.format(result_view_dir, batch_idx), gt_img*255)
            render_img = render_np[0, :, :, :].transpose(1,2,0)
            cv2.imwrite('{}/test_render_{}.png'.format(result_view_dir, batch_idx), render_img*255)
            plot.imsave('{}/test_depth_gt_{}.png'.format(result_view_dir, batch_idx), depth_img, cmap="jet")
            plot.imsave('{}/test_depth_pred_{}.png'.format(result_view_dir, batch_idx), depth_pred_img, cmap="jet")
    return output3


class AverageMeter(object):
    """Computes and stores the average and current value"""
    def __init__(self):
        self.reset()

    def reset(self):
        self.val = 0
        self.avg = 0
        self.sum = 0
        self.count = 0

    def update(self, val, n=1):
        self.val = val
        self.sum += val * n
        self.count += n
        self.avg = self.sum / self.count

    def to_dict(self):
        return {'val' : self.val,
            'sum' : self.sum,
            'count' : self.count,
            'avg' : self.avg}

    def from_dict(self, meter_dict):
        self.val = meter_dict['val']
        self.sum = meter_dict['sum']
        self.count = meter_dict['count']
        self.avg = meter_dict['avg']

abs_rel_error_meter = AverageMeter()
sq_rel_error_meter = AverageMeter()
lin_rms_sq_error_meter = AverageMeter()
log_rms_sq_error_meter = AverageMeter()
d1_inlier_meter = AverageMeter()
d2_inlier_meter = AverageMeter()
d3_inlier_meter = AverageMeter()

abs_rel_error_meter_coarse = AverageMeter()
sq_rel_error_meter_coarse = AverageMeter()
lin_rms_sq_error_meter_coarse = AverageMeter()
log_rms_sq_error_meter_coarse = AverageMeter()
d1_inlier_meter_coarse = AverageMeter()
d2_inlier_meter_coarse = AverageMeter()
d3_inlier_meter_coarse = AverageMeter()

def compute_eval_metrics2(output, gt, depth_mask):
        '''
        Computes metrics used to evaluate the model
        '''
        depth_pred = output
        gt_depth = gt

        N = depth_mask.sum()

        # Align the prediction scales via median
        median_scaling_factor = gt_depth[depth_mask>0].median() / depth_pred[depth_mask>0].median()
        depth_pred *= median_scaling_factor

        abs_rel = abs_rel_error(depth_pred, gt_depth, depth_mask)
        sq_rel = sq_rel_error(depth_pred, gt_depth, depth_mask)
        rms_sq_lin = lin_rms_sq_error(depth_pred, gt_depth, depth_mask)
        rms_sq_log = log_rms_sq_error(depth_pred, gt_depth, depth_mask)
        d1 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=1)
        d2 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=2)
        d3 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=3)
        
        abs_rel_error_meter.update(abs_rel, N)
        sq_rel_error_meter.update(sq_rel, N)
        lin_rms_sq_error_meter.update(rms_sq_lin, N)
        log_rms_sq_error_meter.update(rms_sq_log, N)
        d1_inlier_meter.update(d1, N)
        d2_inlier_meter.update(d2, N)
        d3_inlier_meter.update(d3, N)
 
def compute_eval_metrics1(output, gt, depth_mask):
        '''
        Computes metrics used to evaluate the model
        '''
        depth_pred = output
        gt_depth = gt

        N = depth_mask.sum()

        # Align the prediction scales via median
        median_scaling_factor = gt_depth[depth_mask>0].median() / depth_pred[depth_mask>0].median()
        depth_pred *= median_scaling_factor

        abs_rel = abs_rel_error(depth_pred, gt_depth, depth_mask)
        sq_rel = sq_rel_error(depth_pred, gt_depth, depth_mask)
        rms_sq_lin = lin_rms_sq_error(depth_pred, gt_depth, depth_mask)
        rms_sq_log = log_rms_sq_error(depth_pred, gt_depth, depth_mask)
        d1 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=1)
        d2 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=2)
        d3 = delta_inlier_ratio(depth_pred, gt_depth, depth_mask, degree=3)
        
        abs_rel_error_meter_coarse.update(abs_rel, N)
        sq_rel_error_meter_coarse.update(sq_rel, N)
        lin_rms_sq_error_meter_coarse.update(rms_sq_lin, N)
        log_rms_sq_error_meter_coarse.update(rms_sq_log, N)
        d1_inlier_meter_coarse.update(d1, N)
        d2_inlier_meter_coarse.update(d2, N)
        d3_inlier_meter_coarse.update(d3, N)

 
# Main Function ---------------------------------------------------------------------------------------------
def main():
    global_step = 0
    global_val = 0
    csv_filename = os.path.join(result_view_dir, 'logs/result_log.csv')
    fields = ['epoch', 'Abs Rel', 'Sq Rel', 'Lin RMSE', 'log RMSE', 'D1', 'D2', 'D3']
    csvfile = open(csv_filename, 'w', newline='')
    with csvfile:
        csvwriter = csv.writer(csvfile)  
        csvwriter.writerow(fields) 
    # Start Training ---------------------------------------------------------
    start_full_time = time.time()
    for epoch in tqdm(range(start_epoch+1, args.epochs+1), desc='Epoch'):
        print('---------------Train Epoch', epoch, '----------------')
        total_train_loss = 0
        first_depth_estimation_loss = 0
        stereo_matching_loss = 0
        #-------------------------------
        first_network.eval()
        stereo_network.train()
        # Train --------------------------------------------------------------------------------------------------
        for batch_idx, (rgb, depth, depth_mask) in tqdm(enumerate(train_dataloader)):
            #get ground truth up-down disparity
            sgrid = S360.grid.create_spherical_grid(512).cuda()
            uvgrid = S360.grid.create_image_grid(512, 256).cuda()
            rgb, depth, depth_mask = rgb.cuda(), depth.cuda(), (depth_mask>0).cuda()
            #first depth estimation
            optimizer1.zero_grad()
            optimizer2.zero_grad()
            coarse_depth_pred = first_network(rgb[:,:3,:,:])
             
            attention_weights = S360.weights.theta_confidence(sgrid)
            # berhu depth loss with coordinates attention weight 
            depth_loss = L.direct.calculate_berhu_loss(coarse_depth_pred, depth,                
               mask=depth_mask, weights=attention_weights)
            left_xyz = S360.cartesian.coords_3d(sgrid, coarse_depth_pred)
            dI_dxyz = S360.derivatives.dV_dxyz(left_xyz)               
            guidance_duv = S360.derivatives.dI_duv(rgb[:,:3,:,:])
            depth_smoothness_loss = L.smoothness.guided_smoothness_loss(
                dI_dxyz, guidance_duv, depth_mask, (1.0 - attention_weights)
                * depth_mask.type(attention_weights.dtype)
            )

            coarse_loss = depth_loss #+ depth_smoothness_loss * 0.1
            
            depth_np = coarse_depth_pred.detach().cpu().numpy()
            if batch_idx % visualize_interval == 0 and batch_idx > 0:
                    #print(depth_np.min(), depth_np.max())
                    depth_coarse_img = depth_np[0, 0, :, :]
                    depth_coarse_img[depth_coarse_img>8] = 0
                    plot.imsave('{}/coarse_depth_{}.png'.format(result_view_dir, batch_idx), depth_coarse_img, cmap="jet")
            
            render = dibr_vertical(coarse_depth_pred.clamp(0.1, 8.0), rgb, uvgrid, sgrid, baseline=baseline)
            output, stereo_loss = train(rgb, render, depth.squeeze(1), depth_mask.squeeze(1), batch_idx)
            #need to balance the first network loss and second stereo matching network loss
            alpha, beta = 0.02, 1
            
            loss = stereo_loss*0.02 + coarse_loss 
            #print(stereo_loss.item(), coarse_loss.item(), vgg_loss.item())
            loss.backward()
            optimizer1.step()
            optimizer2.step()
            #coarse_loss = torch.tensor(0)
            total_train_loss += loss.item()
            first_depth_estimation_loss += coarse_loss.item() 
            stereo_matching_loss += stereo_loss.item() 
            global_step += 1
            if batch_idx % 20 == 0 and batch_idx > 0:
                print('[Epoch %d--Iter %d]total loss %.4f, coarse loss %.4f, stereo loss %.4f' % 
                (epoch, batch_idx, total_train_loss/(batch_idx+1), first_depth_estimation_loss/(batch_idx+1), stereo_matching_loss/(batch_idx+1)))
            writer.add_scalar('first network depth loss', coarse_loss, global_step)
            writer.add_scalar('stereo matching loss', stereo_loss, global_step)
            writer.add_scalar('total loss', loss, global_step) # tensorboardX for iter
        writer.add_scalar('train loss epoch',total_train_loss/len(train_dataloader),epoch) # tensorboardX for epoch
        writer.add_scalar('stereo loss epoch',stereo_matching_loss/len(train_dataloader),epoch)
        writer.add_scalar('coarse loss epoch',first_depth_estimation_loss/len(train_dataloader),epoch)
        #---------------------------------------------------------------------------------------------------------
        scheduler1.step()
        scheduler2.step()

        if epoch % 10 == 0:
            epochfilename =  args.visualize_path + '/checkpoints/checkpoint_' + str(epoch) + '.tar'  
            torch.save({
                    'state_dict1': first_network.state_dict(),
                    'state_dict2': stereo_network.state_dict(),
                    }, epochfilename)
        latestfilename =  args.visualize_path + '/checkpoints/checkpoint_latest.tar'  
        torch.save({
                'state_dict1': first_network.state_dict(),
                'state_dict2': stereo_network.state_dict(),
                }, latestfilename)
        #-----------------------------------------------------------------------------

        # Valid ----------------------------------------------------------------------------------------------------
        total_val_loss = 0
        total_val_crop_rmse = 0
        print('-------------Validate Epoch', epoch, '-----------')
        first_network.eval()
        stereo_network.eval()
        for batch_idx, (rgb, depth, depth_mask) in tqdm(enumerate(val_dataloader)):
            sgrid = S360.grid.create_spherical_grid(512).cuda()
            uvgrid = S360.grid.create_image_grid(512, 256).cuda()
            rgb, depth, depth_mask = rgb.cuda(), depth.cuda(), (depth_mask>0).cuda()
            with torch.no_grad():
                coarse_depth_pred = first_network(rgb[:,:3,:,:])
            
            #num_render_view = len(baseline)

            render = dibr_vertical(coarse_depth_pred.clamp(0.1, 8.0), rgb, uvgrid, sgrid, baseline=baseline)

            depth_np = coarse_depth_pred.detach().cpu().numpy()
            if batch_idx % 20 == 0 and batch_idx > 0:
                    depth_coarse_img = depth_np[0, 0, :, :]
                    depth_coarse_img[depth_coarse_img>8] = 0
                    plot.imsave('{}/test_coarse_depth_{}.png'.format(result_view_dir, batch_idx), depth_coarse_img, cmap="jet")

            val_output = val(rgb, render, depth.squeeze(1), depth_mask.squeeze(1), batch_idx)
            compute_eval_metrics1(coarse_depth_pred, depth, depth_mask)
            compute_eval_metrics2(val_output, depth.squeeze(1), depth_mask.squeeze(1))
	        #-------------------------------------------------------------
            # Loss ---------------------------------
            #total_val_loss += val_loss
            #---------------------------------------
            # Step ------
            global_val+=1
            #------------
        #writer.add_scalar('total validation loss',total_val_loss/(len(val_dataloader)),epoch) #tensorboardX for validation in epoch        
        #writer.add_scalar('total validation crop 26 depth rmse',total_val_crop_rmse/(len(val_dataloader)),epoch) #tensorboardX rmse for validation in epoch
        print('Epoch: {}\n'
        '  Avg. Abs. Rel. Error: coarse {:.4f}, final {:.4f}\n'
        '  Avg. Sq. Rel. Error: coarse {:.4f}, final {:.4f}\n'
        '  Avg. Lin. RMS Error: coarse {:.4f}, final {:.4f}\n'
        '  Avg. Log RMS Error: coarse {:.4f}, final {:.4f}\n'
        '  Inlier D1: coarse {:.4f}, final {:.4f}\n'
        '  Inlier D2: coarse {:.4f}, final {:.4f}\n'
        '  Inlier D3: coarse {:.4f}, final {:.4f}\n\n'.format(
        epoch, 
        abs_rel_error_meter_coarse.avg,
        abs_rel_error_meter.avg,
        sq_rel_error_meter_coarse.avg,
        sq_rel_error_meter.avg,
        math.sqrt(lin_rms_sq_error_meter_coarse.avg),
        math.sqrt(lin_rms_sq_error_meter.avg),
        math.sqrt(log_rms_sq_error_meter_coarse.avg),
        math.sqrt(log_rms_sq_error_meter.avg),
        d1_inlier_meter_coarse.avg,
        d1_inlier_meter.avg,
        d2_inlier_meter_coarse.avg,
        d2_inlier_meter.avg,
        d3_inlier_meter_coarse.avg,
        d3_inlier_meter.avg))

        row = [epoch, '{:.4f}'.format(abs_rel_error_meter.avg.item()), 
            '{:.4f}'.format(sq_rel_error_meter.avg.item()), 
            '{:.4f}'.format(torch.sqrt(lin_rms_sq_error_meter.avg).item()),
            '{:.4f}'.format(torch.sqrt(log_rms_sq_error_meter.avg).item()), 
            '{:.4f}'.format(d1_inlier_meter.avg.item()), 
            '{:.4f}'.format(d2_inlier_meter.avg.item()), 
            '{:.4f}'.format(d3_inlier_meter.avg.item())]
        with open(csv_filename, 'a', newline='') as csvfile:
            csvwriter = csv.writer(csvfile)  
            csvwriter.writerow(row)


    writer.close()
    # End Training
    print("Training Ended hahahaha!!!")
    print('full training time = %.2f HR' %((time.time() - start_full_time)/3600))
#----------------------------------------------------------------------------------

if __name__ == '__main__':
    main()