Assume that I have a camera pointing in a specific direction. I know the Euclidean distance (Real world distance) of the camera to a fixed point, X (mm). Using YOLOv3 I have detected all the occurences of an object and have the bounding boxes and their centers. I also am able to determine the Euclidean distance of the camera to all bounding box centers. I have the availability to the following measures:

focal length,
image height (px), ## Photo
image width (px),  ## Photo
Real object height (mm),
Real object width (mm),
sensor height (mm),
sensor width (mm),
pixel size ($mu$m),
Field of view (vertical),
Field of view (horizontal)

and the following calculated measures:

0. The pixel-coordinates of every single bounding box center.
1. Euclidean distance from camera to all instances of a specific and detected object 
2. object height and width (in pixels) # used to calculate 1.
3. Angle made by between every pair of bounding box centers relative a fixed point on the image (actually, an epipole determined by the use of a second image)

Now, what I would like to do is to calculate the Euclidean distance (Real world distance) between the objects in the photo. I naively thought that knowing the pixel distance and the pixel size woul be enough but apparently I was mistaken.

Any insight would be greatly appreciated.




Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Browse other questions tagged or ask your own question.