In this paper we present the design, implementation and preliminary tests of a high depth underwater ROV inspection system. The ROV is equipped with a sensor-based system in order to characterize the water quality of marine ecosystems, and a vision-based system to identify the marine species. The vision system includes an adaptive color correction algorithm and a tracking algorithm that improves in real time the tracking of living species during inspection. A prototype was implemented to validate the functionality of the ROV operation and the vision system algorithms.