RetinaNet模型在DDSM数据集的应用问题（3）

数据处理

我确实按照昨天想的把图片resize成了1000*1000大小 bounding box也重新计算了。
以下是修改以后的matlab代码。

%% ==================================================================%
%     OPEN AND VIEW PNG FORMAT DDSM IMAGES WITH THEIR ANNOTATIONS 
%                        -ope---(MAIN FILE)----
% ===================================================================%
%           Author - Anmol Sharma (Undergraduate Engineering Student)
%      Affiliation - DAV Institute of Engineering & Technology
%      Supervisers - Dr. Jayasree Chakraborty* and Dr. Abhishek Midya#
%                    *Research Fellow, Memorial Sloan Kettering Cancer Center
%                    #Assistant Professor, NIT Silchar
%      Description - The code is used to view the PNG format DDSM images
%                    created using the openDDSMLJPEG1AndConvertToPNG.m 
%                    script. This script opens the PNG file, and then also
%                    opens the corresponding OVERLAY file to get the boundary
%                    information of the mass present in that particular 
%                    mammogram.
%          License - Copyright (C) 2015  Anmol Sharma
%
%                    This program is free software: you can redistribute it
%                    and/or modify it under the terms of the GNU General 
%                    Public License as published by the Free Software 
%                    Foundation, either version 2 of the License, or (at
%                    your option) any later version.
% 
%                    This program is distributed in the hope that it will 
%                    be useful, but WITHOUT ANY WARRANTY; without even the 
%                    implied warranty of MERCHANTABILITY or FITNESS FOR A 
%                    PARTICULAR PURPOSE.  See the GNU General Public License
%                    for more details.
% 
%                    You should have received a copy of the GNU General 
%                    Public License along with this programmax_y.  If not, 
%                    see <http://www.gnu.org/licenses/>. 
%===================================================================%
clear all
clc
%% SET THESE PATHS FIRST!
% Needed to load PNG files
directoryOfDDSMPNG = 'D:\cancers_jpg_test\';
directoryOfDDSMPNG_new = 'D:\cancers_jpg_test_new\';
disp(directoryOfDDSMPNG)
% Needed to load OVERLAY FIles
directoryOfDDSM = 'D:\OVERLAY_CC_test\';
 
 
 
imageOutputFileFormat = '*.jpg'; % Notice the STAR dot. Can be *.tif, *.jpg...
filenames = dir(strcat([directoryOfDDSMPNG, imageOutputFileFormat]));

disp(filenames)
for i = 1:length(filenames)
    disp( filenames(i).name((1:(length(filenames(i).name) - 5))))
    
    overlayName = strcat([directoryOfDDSM, filenames(i).name((1:(length(filenames(i).name) - 4))), '.OVERLAY']);
    jpgName_new = strcat([directoryOfDDSMPNG_new, filenames(i).name((1:(length(filenames(i).name) - 4))), '.jpg']);
    JPGName = strcat([filenames(i).name((1:(length(filenames(i).name) - 4))), '.jpg']);
 
    [num]=count_overlay(overlayName)
   
    for j = 1:num
        [bnd_c,bnd_r] = readBoundary(overlayName, j);
        %bnd_c²¡ÔîÇøÓòµÄËùÓкá×ø±ê£¬bnd_r²¡ÔîÇøÓòµÄËùÓÐÁÐ×ø±ê
       % disp(bnd_c)
        min_x=min(bnd_c)
        max_x=max(bnd_c)
        width=max_x-min_x
        min_y=min(bnd_r)
        max_y=max(bnd_r)
        height=max_y-min_y
        left_x=min_x
        left_y=min_y
  
        image = imread(strcat(directoryOfDDSMPNG, filenames(i).name));
        
        [heightOfImage, widthOfImage] = size(image);     
        B = imresize(image,[1000 1000])
        [heightOfImage1, widthOfImage1] = size(B)
       % temp_mask = poly2mask(bnd_c,bnd_r, 1000,1000);
        imshow(B,[])
       % PATH=
       
        imwrite(B,jpgName_new)
        hold on
        scale_w=1000/widthOfImage
        scale_y=1000/heightOfImage
        A = [left_x*scale_w,left_y*scale_y,width*scale_w,height*scale_y]
        write(JPGName,A)
 
 %       rectangle('Position',A) 
         rectangle('Position',A,'EdgeColor','r')
         plot(bnd_c*scale_w, bnd_r*scale_y, '-r'); 
         pause;
        close all;
    end
   
 
end

我真的很想哭，看到结果和昨天一样，我真的想打人了。我又不知道能和谁去讨论，只能自己在这里写。

• 一共929张图片。这里只是跑了单类别，但是肺炎的程序是用（正常，病灶）两个类别。

我不知道我到底哪里出了问题
1.数据量少？
2.数据问题？—解决办法：

再用retinaNet 跑一遍肺炎的数据集

• 把rsna-肺炎的代码再跑一遍

之后就报错，这个错误我在以前遇到过，应该是版本问题不匹配，后来我也忘了怎么解决的了，不过我现在想要用我的代码跑一遍肺炎的数据集。
实验室要关门了。。。
如果这个结果可以跑通就证明我的代码没有问题，还是数据的问题！！！
如果这个跑不通，就证明我的代码有问题！！

下面的这个bug是用原来的author的codes work的，可能是因为version的问题，不过后来我似乎solve了。。。我实在想不起来我怎么做到的了，，好像以前的blog中record了。。

---

Traceback (most recent call last):

File “train.py”, line 511, in

main()

File “train.py”, line 496, in main

callbacks=callbacks,

File “/home/ld/anaconda3/lib/python3.6/site-packages/keras/legacy/interfaces.py”, line 91, in wrapper

return func(*args, **kwargs)

File “/home/ld/anaconda3/lib/python3.6/site-packages/keras/engine/training.py”, line 1418, in fit_generator

initial_epoch=initial_epoch)

File “/home/ld/anaconda3/lib/python3.6/site-packages/keras/engine/training_generator.py”, line 251, in fit_generator

callbacks.on_epoch_end(epoch, epoch_logs)

File “/home/ld/anaconda3/lib/python3.6/site-packages/keras/callbacks.py”, line 79, in on_epoch_end

callback.on_epoch_end(epoch, logs)

File “keras-retinanet/keras_retinanet/callbacks/common.py”, line 31, in on_epoch_end

self.callback.on_epoch_end(epoch, logs=logs)

File “keras-retinanet/keras_retinanet/callbacks/eval_rsna.py”, line 72, in on_epoch_end

self.model,

File “keras-retinanet/keras_retinanet/utils/eval_rsna.py”, line 360, in evaluate

annotations          = all_annotations[i][label]

IndexError: index 0 is out of bounds for axis 0 with size 0

---

好难过_{真的心都碎了}真的

觉得自己一个人一直扛着所有的事情，就连心里想说的，也不可以和bf说，似乎我们不是一个世界，我也不想过多打扰他的世界，他的世界太过于简单。
我只能沉迷于代码中，其实?
也算是好事吧,否则我一天胡思乱想的，抗住这个压力还蛮好的，我说实话还蛮享受现在的日子的。

回去就想睡觉，看一会《how to change your mind》然后明天又是活力满满的一天~~~~

加油