generate_test_file

############################################################################
# Generates test data by loading the trained model
#                             April 2020
#          Ravi Konkimalla, Neil Rodrigues | University of Pennsylvania
############################################################################


import os
import time
import numpy
import torch
import math
import pdb

import sys
import numpy as np
import torch.nn as nn
from PIL import Image, ImageOps
import torchvision.models as models

from torch.optim import SGD, Adam
from torch.autograd import Variable

from dataloader import Squeeze_Seg
from torch.utils.data import DataLoader
from torchvision.transforms import ToTensor, ToPILImage, Resize
import matplotlib.pyplot as plt

from utils.util_iou_eval import iouEval


from dataloader import Squeeze_Seg
from infer_config import *

sys.path.append(os.path.join(ARGS_ROOT,'models',ARGS_MODEL_NAME+'/'))
print(ARGS_MODEL_NAME)
module = __import__(ARGS_MODEL_NAME)
Network = getattr(module,"Net")

from utils.util_iou_eval import iouEval, getColorEntry
from utils.calculate_weights import load_class_weights

device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')

def load_my_state_dict(model, state_dict):
    
    own_state = model.state_dict()
    
    for name, param in state_dict.items():
        if name not in own_state:
            print("[weight not copied for %s]"%(name)) 
            continue
        own_state[name].copy_(param)
    return model

def test(model,output_dir):
    print('Total Number of classes is {}'.format(NUM_CLASSES))

    dataset_val = Squeeze_Seg(ROOT_DIR,'08',ARGS_INPUT_TYPE_1,ARGS_INPUT_TYPE_2)

    loader_val = DataLoader(
        dataset_val,
        num_workers = ARGS_NUM_WORKERS,
        batch_size = ARGS_VAL_BATCH_SIZE,
        shuffle = False)

    weight = load_class_weights()

    iouEvalVal = iouEval(NUM_CLASSES)
    labels = np.array([(0,0,0),(255,0,0),(0,255,0),(0,0,255)] )
    model.eval()
    total_time=0
    iters = 00
    for step, (image,image_2,mask,label) in enumerate(loader_val):
            


        if ARGS_CUDA:
            image = image.cuda()
            image_2 = image_2.cuda()
            label = label.cuda()
            mask = mask.cuda()

        image = Variable(image)
        label = Variable(label)
        mask = Variable(mask)
        image_2 = Variable(image_2)
        print(image.shape)
        output = model(image,image_2,mask)

        
        label_out = output[0].max(0)[1].data
        image = image[0,:,:,:].permute(1,2,0)
        image = image[:,:,:3]
        image = torch.cat((image,label_out.view(64,512,1).float()),axis = 2)

        output_path = output_dir + str(iters)
        np.save(output_path, image.cpu().numpy())
        print(iters)
        iters += 1
        
        if iters >= 400:
            break
    



if __name__ == "__main__":
    
    print(device)
    model_type = 'res_fcn'
    models_pth = 'E:/project_522/train_models/'
    model_pth = models_pth + model_type + '.pth'
    model = Network(data_dict)
    torch.set_num_threads(ARGS_NUM_WORKERS)
    model.load_state_dict(torch.load(model_pth))
    data_type = 0
    seq_str_num = '08'
    pgm_dir = 'E:/project_522/Kitti/pgm_output/' + seq_str_num +'/'
    output_dir = 'E:/project_522/Kitti/pgm_test/' + model_type + '/updated' + '/'
    if not os.path.exists(output_dir):
      os.makedirs(output_dir)
    test(model,output_dir)