🚲Annotating Bicycle

For an example purpose we will be annotating bicycle image.

1. Configuration Setup

Step 1: Task Information

At first, we will provide the task information. The annotated image will be used for segmentation purpose so image segmentation task is selected here.

Then we click next to navigate the next tab.

Step 2: Annotation Configuration

Now, we will pick the tools that we will require for annotation purpose and also provide the labels name that will be used during annotation.

Then we click next to navigate the image tab.

Step 3: Upload Image

Now, we will upload the image that will be used for annotation purpose.

After uploading the image we can click the open lab and start the annotation.

2. Annotating Image

Step 1: Tools Selection

From the tools selection, we can select the tools that we needed. For example purpose circle tool is selected.

Step 2: Draw and Fill the information

The selected tool is draw on top of the region of interest (ROI). For example purpose we will be using it on back tyre.

A popup box will appear. Where we can select the labels from dropdown and provide comments too.

After entering the information, we can click the checkmark.

Now, if we look at the right side bar labels section, the tyre label is highlighted and has a count of 1. It means that current annotated image has a single tyre label with it. The regions section has also a new region added whereas the history section has numerous count. One can lock, unlock, delete, hide or resize label from regions section and can revert changes from history section. We can add as many labels we want.

Sidebar

3. Downloading Changes

Step 1: Save

Save the current change by clicking the Save button from Controls menu.

This will save the current state.

Step 2: Download

To download the configuration, annotated or masked image one can click the download menu button from Controls menu. It will expand the menu and show respective buttons.

After clicking configuration menu, a json file with configuration is downloaded. Here's the content of the downloaded JSON file.

{
   "bianchi.jpeg":{
      "configuration":[
         {
            "image-name":"bianchi.jpeg",
            "regions":[
               {
                  "region-id":"7071021156029009",
                  "image-src":"http://annotate-lab.onrender.com/uploads/bianchi.jpeg",
                  "class":"Tyre",
                  "comment":"Back Tyre",
                  "tags":"",
                  "rx":[
                     0.05460252542288541
                  ],
                  "ry":[
                     0.3497771621890547
                  ],
                  "rw":[
                     0.35672614208955233
                  ],
                  "rh":[
                     0.5774993194029854
                  ]
               }
            ],
            "color-map":{
               "Tyre":[
                  244,
                  67,
                  54
               ]
            }
         }
      ]
   }
}

If we download the annotated and masked image, we will get following images respectively.