Different countries suffer from the problem of detecting and locating the unexploded ordnance (UXO) after military events. Such a problem has a great impact on economic development and future plans. Egypt is one of those countries, where more than 23 million UXOs are expected to cover a wide part of its northwestern desert. Detection of UXO becomes a persistent target for many geophysical research teams. Currently, unconventional near-surface flight technologies, such as quad/hexa-copters instead of regular land surveys, are used for safety reasons in the acquisition phase. In the interpretation phase, the analytical methods played challenging role to solve such ill-posed problems using the regularization tools especially those based on Tikhonov formula. Those methods still have limitations when dealing with highly ill-posed problems. Fortunately, neither noise nor ill-posedness affects the quality of the solution produced by artificial intelligence based techniques. Hence, the application of the hybrid regularize/artificial intelligence based technique would present a promising choice. In this work, we implement the Kaczmarz-Bat technique, which holds the power of the regularizing analytical technique as an initial good solution and the benefits of artificial intelligent technique, represented in bat method to enhance the solution. The gradient magnetic field over UXO objects was sliced into profiles and solved using the proposed approach as 2D earth models. The solutions are stacked together in a 3D perspective model illustrating the depth and location of the targets.