Support PASCAL-VOC-like export format

[corenel]

I notice that placeholders in Create new Exportformat page isn't sufficient to create a PASCAL-VOC-like export format.

For example, PASCAL VOC format looks like this:

<annotation>
    <folder>demo</folder>
    <filename>demo3.jpg</filename>
    <path>/home/yuthon/Workspace/demo/demo3.jpg</path>
    <source>
        <database>Unknown</database>
    </source>
    <size>
        <width>799</width>
        <height>401</height>
        <depth>3</depth>
    </size>
    <segmented>0</segmented>
    <object>
        <name>class-1</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>487</xmin>
            <ymin>107</ymin>
            <xmax>790</xmax>
            <ymax>318</ymax>
        </bndbox>
    </object>
    <object>
        <name>type-f</name>
        <pose>Unspecified</pose>
        <truncated>0</truncated>
        <difficult>0</difficult>
        <bndbox>
            <xmin>141</xmin>
            <ymin>68</ymin>
            <xmax>363</xmax>
            <ymax>334</ymax>
        </bndbox>
    </object>
</annotation>

The filename and path part isn't shown on given placeholders, thus it's impossible to create a PASCAL-VOC-like export format.

I'll appreciate it if you could supply more placeholders, or directly support built-in PASCAL-VOC-like export format.

[NFiedler]

Thank you very much for your contribution and your ideas! I don't want to directly implement any format because i want to keep the flexibility at the highest possible level. So i'd rather add functionality to the export creation itself. As far as i see, the path is the only problem here. The filename has the %%image placeholder. Because the path to the file depends on the system i'd recommend using a shell variable (e.g. export IMAGE_SET_PATH='/home/nfiedler/Bit-Bots/imagetagger') and saving the images in a directory in that path named after the imageset. Then you could use the export format as follows:

<folder>%%imageset</folder>
<filename>%%image</filename>
<path>$IMAGE_SET_PATH/%%imageset/%%image</path>

Therefore the %%imageset placeholder is needed in the annotation format. This is missing right now.

[corenel]

Thanks for your help! I'v tried this:

• for Base format

  <annotation>
  <folder>%%imageset</folder>
  <location>%%setlocation</location>
  <team>%%team</eram>
  <source>
      <database>Unknown</database>
  </source>
  <segmented>0</segmented>
  %%content
  </annotation>

• for Annotation format

  <filename>%%image</filename>
  <path>$IMAGE_SET_ROOT/%%imageset/%%image</path>
  <size>
      <width>%%imagewidth</width>
      <height>%%imageheight</height>
      <depth>%imagedepth</depth>
  </size>
  <object>
      <name>%%type</name>
      <noobject>0</noobject>
      <pose>Unspecified</pose>
      <truncated>0</truncated>
      <difficult>0</difficult>
      <verified>%%veriamount</verified>
      <bndbox>
          <xmin>%%x1</xmin>
          <ymin>%%y1</ymin>
          <xmax>%%x2</xmax>
          <ymax>%%y2</ymax>
      </bndbox>
  </object>

• for Not in image format

  <filename>%%image</filename>
  <path>$IMAGE_SET_ROOT/%%imageset/%%image</path>
  <size>
      <width>%%imagewidth</width>
      <height>%%imageheight</height>
      <depth>%imagedepth</depth>
  </size>
  
  <object>
      <name>%%type</name>
      <noobject>1</noobject>
      <pose>Unspecified</pose>
      <truncated>0</truncated>
      <difficult>0</difficult>
      <verified>%%veriamount</verified>
  </object>

  But there comes two problems:

• You don't supply the placeholders for width, height and depth of a single image file;

• It seems that current exporting util is designed to export all annotations of all images into single file. I'm wondering if it's convenient for you to add a feature to export all annotations of each image into a single file and then compress them up.

Of course, if you're focusing on add other useful features which is much more important, I can also write a tiny Python script to convert the current all-in-one annotation file into what I desired. That's quite easy.

Thanks for your great contribution on building such an excellent image labelling tool!

[NFiedler]

Now i understand the problem. I think two features are needed to realize this functionality:

• aggregation of annotations by image
  • an additional image form is required (base, image, annotation, not in image)
• exporting into multiple files (one for every image)

The aggregation of annotations by image would make it very easy to split the file locally until the splitting feature is implemented.

[corenel]

Thanks for your quick reply! Then I'll write my own scripts for temporary use and wait for the splitting feature.

[NFiedler]

Meanwhile the aggregation by image is possible and the additional placeholders are implemented. I wouldn't like to have the splitting feature on our hosted version, because it leads to a high server load (creation and zipping of thousands of files for large imagesets), and is easily achievable with a simple script on the client side.