Estimating the 3D location of an object from a single camera is an important topic in the field of computer
vision and computer graphics. The problem of transforming a 2D point from a single image to a 3D point is
difficult unless it is located in the same plane, using an RGBD camera or at least using two images. Our study
aims to discuss the perspective grid concept for estimating the 3D location of chair objects accurately on the
floor surface using a single image from a cellphone camera. Our proposal to predict the 3D location of a
single image from a camera goes through three experimental stages. First, setting the nine locations and four
actual object orientations over the floor pattern in the room to get the bounding box position by utilizing the
object detection pre-trained model. The second stage is the development of the perspective grid algorithm as
a transformation of 2D points in the projected image to 3D points on the floor plane. The third stage is
predicting the optimal object location in the image from the lower left and lower right bounding box positions
(assumed to be on the floor surface where the z value is 0) with a perspective grid approach, to be projected
into a 3D location prediction value on the floor plane. Then we evaluate the calculation of the deviation of
the average prediction error from nine actual object locations, each object location, and each object
orientation. The average error deviation result is 6.47 centimeters. This shows that the results are quite
accurate compared to the dimensions of the object and the area of the room