The integration of robotics into daily life applications is growing every day. Unmanned Aerial Vehicles (UAVs) are being extensively used for inspecting indoors, underground, surveying disastersites, and tactical Search-and Rescue missions. As these applications are mainly dependent on GPS signals, awareness increased of its two major short comingsI) weak-to no GPS signal sat these locations, andII) easeat which the agents could be intentionally blocked from GPS access or communication signal satmilitary or criticalsites. Although this technology is still evolving, the movement toward developing algorithms that allow the operation of UAVs swarm is rising; to allow sharing of complicated tasks amongst the agents reducing the role of human operators. This paper discusses are liable solution allowing aswarm of UAVs to traverse terrains in GPS- denied environments, using visual odometry to geolocalize. The proposed localization methodology comprises two stages: database preparation for the environment and on board inference forreal time localization and navigation. The experimental results show that the propose dimage based localization succeeded in localizing the quadcopter and sharing its position with other quadcopters in the swarm formation. Additionally, the proposed swarm formation is able to control the leader and follower quadcopters to follow aprede fined path with pre-set relative distance between the leader and follower.