TCSVT-VCSNet/test_VCSNet.m at master · wzhshi/TCSVT-VCSNet · 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
% clear all; close all;
addpath('./utilities');
run('D:\AcademicSoftware\matconvnet-1.0-beta25\matlab\vl_setupnn.m') ;

total_seq_num = 6;
total_frame_num = 17; % to be tuned
GOP_size = 8; % to be tuned

key_subrate = 0.5;
subrate = 0.1;  % uniform Sampling Rate

block_size = 32; % Block Size for BCS

im = zeros(288,352,17,'single');

imgPSNR  = zeros(total_seq_num, total_frame_num);
imgSSIM  = zeros(total_seq_num, total_frame_num);

savet = 1;

 net = load(['./models/net' num2str(subrate) '.mat']);
 net = dagnn.DagNN.loadobj(net.net);

showResult  = 0;
useGPU      = 0;
pauseTime   = 0;
if useGPU
    net.move('gpu');
end

for kk = 1:6   % for each sequence   modified 1:6
    switch kk
         case 1
            sequence_name = 'akiyo_cif';
        case 2
            sequence_name = 'coastguard_cif';
        case 3
            sequence_name = 'Foreman_cif';
        case 4
            sequence_name = 'Mother_daughter_cif';
        case 5
            sequence_name = 'Paris_cif';
        case 6
            sequence_name = 'Silent_cif';
    end

    Opts.sequence_name = sequence_name;

    for i = 1 : total_frame_num    % for each frame of this sequence
        filename = ['./Sequences/' sequence_name '/' sequence_name '_' num2str(i) '.png'];
        im(:,:,i) = im2single(imread(filename));
    end

    if useGPU
        im = gpuArray(im);
    end

    net.eval({'input',im(:,:,1:9)});
    Recon1 = zeros(288,352,8,'like',im);

    Recon1(:,:,1) = net.vars(net.getVarIndex('keyFdr_pred')).value;
    Recon1(:,:,2) = net.vars(net.getVarIndex('nkeyF1dr_pred')).value;
    Recon1(:,:,3) = net.vars(net.getVarIndex('nkeyF2dr_pred')).value;
    Recon1(:,:,4) = net.vars(net.getVarIndex('nkeyF3dr_pred')).value;
    Recon1(:,:,5) = net.vars(net.getVarIndex('nkeyF4dr_pred')).value;
    Recon1(:,:,6) = net.vars(net.getVarIndex('nkeyF5dr_pred')).value;
    Recon1(:,:,7) = net.vars(net.getVarIndex('nkeyF6dr_pred')).value;
    Recon1(:,:,8) = net.vars(net.getVarIndex('nkeyF7dr_pred')).value;

    if useGPU
        Recon1 = gather(Recon1);
    end

    for j=1:8
        [psnr,ssim] = Cal_PSNRSSIM(im2uint8(im(:,:,j)),im2uint8(Recon1(:,:,j)),0,0);
        imgPSNR(kk,j) = psnr;
        imgSSIM(kk,j) = ssim;
        if savet
            if j == 1
                imwrite(im2uint8(Recon1(:,:,j)),['./VCSNet-gop8-doubleRef-results/' num2str(subrate),'/' sequence_name, '/' sequence_name,'_',num2str(j),'_rate_',num2str(key_subrate),...
                    '_PSNR_',num2str(psnr),'_SSIM_',num2str(ssim),'.png']);
            else
                imwrite(im2uint8(Recon1(:,:,j)),['./VCSNet-gop8-doubleRef-results/' num2str(subrate),'/' sequence_name, '/' sequence_name,'_',num2str(j),'_rate_',num2str(subrate),...
                    '_PSNR_',num2str(psnr),'_SSIM_',num2str(ssim),'.png']);
            end
        end
    end

    net.eval({'input',im(:,:,9:17)});
    Recon2 = zeros(288,352,9,'like',im);

    Recon2(:,:,1) = net.vars(net.getVarIndex('keyFdr_pred')).value;
    Recon2(:,:,2) = net.vars(net.getVarIndex('nkeyF1dr_pred')).value;
    Recon2(:,:,3) = net.vars(net.getVarIndex('nkeyF2dr_pred')).value;
    Recon2(:,:,4) = net.vars(net.getVarIndex('nkeyF3dr_pred')).value;
    Recon2(:,:,5) = net.vars(net.getVarIndex('nkeyF4dr_pred')).value;
    Recon2(:,:,6) = net.vars(net.getVarIndex('nkeyF5dr_pred')).value;
    Recon2(:,:,7) = net.vars(net.getVarIndex('nkeyF6dr_pred')).value;
    Recon2(:,:,8) = net.vars(net.getVarIndex('nkeyF7dr_pred')).value;
    Recon2(:,:,9) = net.vars(net.getVarIndex('keyF9dr_pred')).value;

    if useGPU
        Recon2 = gather(Recon2);
    end

    for j=9:17
        [psnr,ssim] = Cal_PSNRSSIM(im2uint8(im(:,:,j)),im2uint8(Recon2(:,:,j-8)),0,0);
        imgPSNR(kk,j) = psnr;
        imgSSIM(kk,j) = ssim;
        if savet
            if j == 9
                imwrite(im2uint8(Recon2(:,:,j-8)),['./VCSNet-gop8-doubleRef-results/' num2str(subrate),'/' sequence_name, '/' sequence_name,'_',num2str(j),'_rate_',num2str(key_subrate),...
                    '_PSNR_',num2str(psnr),'_SSIM_',num2str(ssim),'.png']);
            else
                imwrite(im2uint8(Recon2(:,:,j-8)),['./VCSNet-gop8-doubleRef-results/' num2str(subrate),'/' sequence_name, '/' sequence_name,'_',num2str(j),'_rate_',num2str(subrate),...
                    '_PSNR_',num2str(psnr),'_SSIM_',num2str(ssim),'.png']);
            end
        end
    end


    disp([mean(imgPSNR(kk,1:16)),mean(imgSSIM(kk,1:16))])

end


if savet
    xlswrite(['./VCSNet-gop8-doubleRef-results/' num2str(subrate) '/Rate_' num2str(subrate) '_PSNR_results_All_sequences.xlsx'],imgPSNR);
    xlswrite(['./VCSNet-gop8-doubleRef-results/' num2str(subrate) '/Rate_' num2str(subrate) '_SSIM_results_All_sequences.xlsx'],imgSSIM);
end