A scientific paper led by PhDs in Image Processing and Robotics from the University of Pecs, Hungary, compared the performances of our ZED cameras with Intel’s Realsense.
Read the full scientific paper Perspectives of Realsense and ZED Depth Sensors for Robotic Vision Applications (Tadic et al., 2022) here.
A series of experiments were conducted in various lighting conditions and with various objects using the D435, D415, ZED, and ZED 2i depth cameras.
In all the examples, the position and the lighting conditions such as the illumination, angle, shadows, etc., of all the depth sensors were the same for a particular example, so that a fair comparison of the results could be made.
In this example below of a street taken from the sidewalk, the ZED and ZED 2i generated a depth map of very good quality with the fence and the car clearly visible, which is not the case for the images captured with the RenseSense cameras. We can also note that the ZED 2i provided shaper details and its color image is more detailed thanks to its wider optics and field of view.
Color images and depth maps of a street taken with four different depth cameras
All the test samples in different conditions allowed the researchers to draw the following conclusions:
In robotic vision applications where high precision and accuracy were mandatory, the ZED [and ZED 2i] depth sensors yielded better results.
In applications where the required objects were very dark and textureless, along with dim illumination and shadows, the ZED 2i depth sensor generated the most reliable and accurate depth map of the real scene.
If you want to generate reliable and accurate depth maps for your Robotic Vision Applications, get the ZED 2i now from our online store!