endToEnd_stereoLocalization.zip

A labelled dataset for 3D object localization. Each example is composed of two 300x300 images (from two fixed cameras) and the target is a vector with 6 entries representing the XYZ coordinates of two points on the screw driver handle. The two points are the uppest and lowest point of the handle on the screw driver axis. This dataset was generated using an industrial robot to get knowledge of the exact position of the object. The XYZ coordinates are expressed in the robot frame, hence the 3D localization model needs to handle the change of frame. A computer vision algorithm was implemented to remove the robot from some of the images. For other cases, we hide the robot physically with a t-shirt wrapped around the effector. The information relative to the situation, as well as the screw driver present on the two images can be found in the 'info.txt' file.