Encoded Contour Information Format Question

[imtiendat0311]

Hi TagLab developer,

Since TagLab not support huge image. So currently my workflow for very big image as follow:

1. I ran the image thru SAM model to get all the segmentation mask
2. From output mask i tried to find the contour of each mask
3. From all of contour i tried to construct a json project file that follow the taglab format
4. Do some more fine edit using Taglab

So my problem currently with this approach is when i using json file i create to use with taglab. All the annotation seem have incorrect scaling ( the shape seem right but it's small and display over top left of canvas ). Could you help me understand on what actually you guy encoded for contour so that i could display properly in term of size and position when inside taglab app

Incorrect Example

Correct Example

[maxcorsini]

Hi,

yes, it seems that the scale is the only problem, not the encoding of the contour. Taking into account the following: the contours are encoded in pixels in the JSON. Each orthoimage has an associated scale that is the size of a pixel in millimiters that the user should provide when the map is added to the project, to obtain measurements on the map. So, pay attention that when you create your JSON file the contours are in pixel coordinates.

Hope it helps.