DEEP LEARNING OBJECT DETECTION APPLICATION TO SURFING WAVE QUALITY

Quantitative monitoring is imperative to the sustainable management of coastal resources. Surfing resources have been both created and degenerated or destroyed by activities in zone. Effective surfing wave quality requires identification tracking breaking part unbroken crest. Remote Camera Systems (RCS) proven their utility being able monitor zone provide almost continuous, high frequency data collection. RCSs lend themselves very well surf breaks which are highly dynamic. The images captured from an RCS a break on west coast Aotearoa New Zealand used train Convolutional Neural Network (CNN) detect points (BP), associated crest orientation relative Still Water Level (SWL) each instance waves image. Model settings image annotations were modified over suite training cases improve model efficacy, was evaluated epoch with mean Average-Precision (mAP; max 1). A mAP 0.794 achieved for BP Crest Point (CP) CNN, 8.634 SWL. ~1.6 M objects across ~1 million images, confidence value all BP-CP detections 0.63 more than 70percent greater 0.5. This enables first automation meaningful monitoring.