Method and system for detecting a three-dimensional object in a two-dimensional image

A system for detecting a three-dimensional object in a two-dimensional (2D) image, comprising:

• a storage medium;
• an image capturing module configured to continuously capture images and to transmit the images to said storage medium;
• and
• a calculating module that is configured to
• access said storage medium to obtain a to-be-detected image that includes a to-be-detected object,
• input a to-be-detected image into an object detection model, and obtain a resulting output of the object detection model that includes a detection depth dataset, the detection depth dataset including a regional subset associated with a 2D bounding box bounding the to-be-detected object and a depth marking subset associated with the regional subset, the depth marking subset including a coordinate set of a detected reference point associated with the to-be-detected object, a length of the to-be-detected object, a first angle defined by a reference line that crosses the detected reference point and that is parallel to an edge of the 2D bounding box and a first line that originates from the detected reference point and that extends along the length of the to-be-detected object, a width of the to-be-detected object, a second angle defined by the reference line and a second line that originates from the detected reference point and that extends along the width of the to-be-detected object, a height of the to-be-detected object, and a third angle defined by the reference line and a third line that originates from the detected reference point and that extends along the height of the to-be-detected object,
• perform a calculation operation based on the detection depth dataset, so as to obtain coordinate sets of a number of points-of-interest each associated with the to-be-detected object in the to-be-detected image, in a three-dimensional (3D) camera centered coordinate system, and
• perform a mapping operation to convert the coordinate sets of the number of points-of-interest in the 3D camera centered coordinate system into a number of coordinate sets, respectively, in a 3D global coordinate system.