Contour segmentation is fast but not so precise
Lightweightness comes with a price. Despite contour segmentation being fast, it isn't so stable and suffers from pattern leaking and other things.

Experiment deep learning approaches with speed in mind
Maybe YOLO (focused on the bounding boxes, not the classification) or Encoder-Decoder (like U-Nets) architectures should be tested to check if it can be better on segmentation (almost certain) but also fast (uncertain), without using GPUs.

• Currently the max resolution is set to FullHD (1920x1080), while corner supports lower resolutions. Video streams higher than 1920x1080 are downscaled to this resolution in real-time;
• The idea here is to unlock higher resolutions and check the overall system performance;
• Also the user should have the opportunity to choose the work resolution.

Today only centimeters are supported
corner must be more well localized and supports other units of measure, like foot, inches, etc. It also must support more metric units subdivisions.

Create conversion methods
This is pretty easy and straightforward and corner should let the users chose their units.

• Currently, on the detected objects, only right angle segments (bounded by the box) are measured;
• All other object segments must be measured too.

Only Ubuntu/Debian based packages (.deb) are generated

Describe the solution you'd like

• It must support most of Linux distros binary packages, like rpm, or other package managers, like snap or flatpak;
• MacOS supported packages must also be included;
• Windows installers is important too.
• Mobile packages like for Android or iOS

GUI for humans
For most people, using an application through a graphical interface is the preferable way to do this and corner doesn't have one.

Solution
An operating system agnostic interface must be created, maybe using opencv-highgui

Alternatives
Built it using qt.

Only one segment ruler
Currently the virtual ruler supports only one segment for calculating some measures.

Make it calculate n segments
A proposed behaviour is:

• User (left) clicks/touch some point of the video and the ruler appears (current behaviour);
• The ruler follows the mouse pointer/touch point and the calculation happens real-time (current behaviour);
• On a new (left) click/touch, the current ruler becomes static and from its endpoint, a new ruler appears (new behaviour);
• It continues until a ruler limit is reached or the user (right) click or double (left) click.
• All the rulers must be preserved until a clear command is issued.