Nowadays, computer vision (CV) is widely used to solve real-world problems, which pose increasingly higher challenges. In this context, the use of omnidirectional video in a growing number of applications, along with fast development of Deep Learning (DL) algorithms for object detection, drives the need for further research to improve existing methods specifically developed for conventional 2D planar video. This work explores DL methods to detect visual objects in omnidirectional images represented onto plane through Equirectangular Projection (ERP). It is shown that the error rate of object detection using existing DL models with ERP images depends on the object spherical location in the image. Then, a new object detection framework is proposed to obtain uniform error rate across the whole spherical image regions.