rd_map_temporal_spatial_denoising_autoencoder/predict.py at main · GuanRunwei/rd_map_temporal_spatial_denoising_autoencoder · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
import numpy as np
import pandas as pd
import torch
import torch.backends.cudnn as cudnn
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader
from tqdm import tqdm
from tqdm import trange
from utils.dataloader import get_dataloader
from nets.nano_sta_decoder import nano_sta
from torch.utils.tensorboard import SummaryWriter
from utils.callbacks import loss_save
import datetime
import matplotlib
matplotlib.use('TkAgg')
import matplotlib.pyplot as plt
import random
import os

def plot_denoised_rd_maps(denoised_rd_maps):
    fig, ax = plt.subplots(3, 7, figsize=(15, 12))
    row_index = 0
    data_index = 1

    for index in range(denoised_rd_maps.shape[0]):
        rd_map_noise = denoised_rd_maps[index]
        ax[row_index, index % 7].imshow(rd_map_noise)
        ax[row_index, index % 7].set_title("Frame {},Noise {}".format(row_index + data_index, index % 7))

        if (index + 1) % 7 == 0 and index != 0:
            row_index += 1

        if row_index == 3:
            break

    plt.show()


def export_rd_map():
    model_path = "models/radar_stda.pth"
    device = 'cuda:0' if torch.cuda.is_available() else 'cpu'
    model = nano_sta(encoder_in_channels=1, encoder_out_channels=128, decoder_out_channels=1).to('cpu')
    model.load_state_dict(torch.load(model_path))
    model.eval()
    final_results = []
    gts = []
    mapts = []

    train_data_read = pd.read_pickle(
        "E:/Big_Datasets/RaDICaL_Denoising/RD_map_log/RD_map_log/temporal_spatial_data.pkl")
    print("======== start export ==========")
    for i in trange(14, 35):
        example_data = train_data_read.loc[i]
        map_t, map_t_1, map_t_2, gt = example_data['t'], example_data['t-1'], example_data['t-2'], \
                                      example_data['gt']

        input_map = torch.from_numpy(np.array(map_t)).unsqueeze(1).unsqueeze(2).type(
            torch.FloatTensor)

        output_map = model(input_map).squeeze(0).squeeze(0)
        final_results.append(output_map.detach().numpy())
        gts.append(gt)
        mapts.append(map_t)

    final_results = np.array(final_results)
    final_results = final_results.reshape(final_results.shape[0], final_results.shape[1]*final_results.shape[2], -1)
    gts = np.array(gts)
    mapts = np.array(mapts)
    print(final_results.shape)
    print(gts.shape)
    print(mapts.shape)
    np.save("cae.npy", final_results)
    np.save("gt_maps.npy", gts)
    np.save("mapts.npy", mapts)

    plot_denoised_rd_maps(final_results)
    print("======== end export ==========")


if __name__ == '__main__':

    model_path = "models/nano_sta.pth"
    device = 'cuda:0' if torch.cuda.is_available() else 'cpu'

    nano_sta_model = nano_sta(encoder_in_channels=1, encoder_out_channels=128, decoder_out_channels=1).to(device)
    nano_sta_model.load_state_dict(torch.load(model_path))
    nano_sta_model.eval()

    train_data_read = pd.read_pickle(
        "E:/Big_Datasets/RaDICaL_Denoising/RD_map_log/RD_map_log/temporal_spatial_data.pkl")
    random_index = random.randint(20000, 23966)
    print("random index:", 0)
    example_data = train_data_read.loc[random_index]
    map_t, map_t_1, map_t_2, gt = example_data['t'], example_data['t-1'], example_data['t-2'], \
                                  example_data['gt']

    np.save("qualitive_rd_map_example.npy", map_t)

    input_map = torch.from_numpy(np.array([map_t, map_t_1, map_t_2])).unsqueeze(1).unsqueeze(2).type(torch.cuda.FloatTensor)
    print(input_map.shape)

    output_map = nano_sta_model(input_map)

    output_map = output_map.squeeze(0).squeeze(0)
    print(output_map.shape)
    output_map = output_map.detach().to('cpu').numpy()


    fig, ax = plt.subplots(1, 5, figsize=(15, 7))
    ax[0].imshow(map_t)
    ax[0].set_title("Frame t interference")

    ax[1].imshow(map_t_1)
    ax[1].set_title("Frame t-1 interference")

    ax[2].imshow(map_t_2)
    ax[2].set_title("Frame t-2 interference")

    ax[3].imshow(gt)
    ax[3].set_title("Frame t gt")

    ax[4].imshow(output_map)
    ax[4].set_title("Frame t decouple")

    plt.imshow(output_map)
    now = datetime.datetime.now()
    plt.savefig(f"images/test_{random_index}_{now.year}{now.month}{now.day}{now.hour}{now.minute}{now.second}.png")
    plt.show()

    export_rd_map()