How to detect multiple objects in image with boundig boxes and make a Matrix with data

Hi, I have an image with several objects demarcated with boundig boxes in a Ground Truth vector (GT). Furthermore, there are two matrices (M1 and M2) that have boundig boxes and demarcate possible objects in the image and specific data of theses objects. I would like to build a matrix resulting from the overlap of the boundig boxes containing the data from the original matrices. The figure illustrates an example of four objects demarcated in green (Ground Truth - GT vector), an M1 matrix containing five elements (boundig boxes and data) and an M2 matrix containing seven elements (boundig boxes and data). The resulting matrix must contain a column corresponding to Ground Truth - GT (0 = no object exists, 1 = object exists) and two columns that must contain the data of the corresponding objects in the image (which overlap each other).

Answer 1

[Image Analyst]

You ask "I would like to build a matrix resulting from the overlap of the boundig boxes containing the data from the original matrices."

See my attached utility to get the overall bounding box from the individual ones. It will overlap them all.

Adapt as needed.

Answer 2

[yanqi liu]

sir，i think may be use YOLO to detect，and then consider compute iou to reduce repeat box