Balancing Robustness and Efficiency in Deep Diver Detection

In this project, we address the inherent difficulties of underwater visual detection by exploiting deep visual object detection models.  

deep_diver_detection

      Figure 1: Schematic diagrams of the deep visual models used for diver detection

 

In addition, we design a CNN-based diver detection model to balance the trade-offs between robustness and efficiency.

proposed_diver_detection

Figure 2: Schematic diagrams of the proposed CNN-based model

 

The proposed model provides considerably faster running time, in addition to achieving detection performances comparable to the state-of-the-art models. For further details and results, please follow this link.  

perf_omp

Figure 3: Detection performances of the proposed model