Python tool that captures the active window, lets you select a Region of Interest (ROI) within that window, and then prints a neatly organized summary of the ROI coordinates (window-relative, absolute, and normalized).
When working on computer vision projects or automation tasks, you often need coordinates from a specific part of the screen or an application window. However, if you share your code with others or deploy it on different screens, the hard-coded coordinates may no longer match.
By using this tool, you can normalize ROIs, making it possible to re-scale them to fit any window or screen size. This saves time, reduces confusion, and helps ensure consistent results across diverse environments.
- Captures the active window and prompts you to select an ROI using an OpenCV UI.
- Outputs three categories of ROI info:
- Window-Relative:
(x, y, w, h)within the window. - Absolute: Screen-based edges and corner coordinates.
- Normalized:
(x, y, w, h)as fractions of the active window size (for easy scaling).
- Window-Relative:
- Supports single-run mode (exits immediately after one ROI capture) or continuous mode (capturing multiple ROIs until you press
Esc). - Configurable capture key (default:
o).
Example Output:
===== ROI SUMMARY =====
Monitor Resolution: 1920 x 1080
Active Window Title: zappybiby/GrabScaledROI: Grab ROI and Normalize Coords - Google Chrome
Active Window Size: 1936 x 1056
ROI (Window-Relative):
Rectangle: x=368, y=468, w=354, h=394
Corners:
top-left = (368, 468)
top-right = (722, 468)
bottom-left = (368, 862)
bottom-right = (722, 862)
ROI (Absolute):
Rectangle: left=360, top=460, right=714, bottom=854
Corners:
top-left = (360, 460)
top-right = (714, 460)
bottom-left = (360, 854)
bottom-right = (714, 854)
ROI (Normalized to Window):
Rectangle: x=0.1901, y=0.4432, w=0.1829, h=0.3731
Corners:
top-left = (0.1901, 0.4432)
top-right = (0.3729, 0.4432)
bottom-left = (0.1901, 0.8163)
bottom-right = (0.3729, 0.8163)
===== END OF ROI INFO =====