Drone-based monitoring protocol and classification: aerial assessment of Hand Gathering activities

The study has clearly demonstrated the potential of Machine and Deep Learning tools for the identification and quantification of hand gathering activity in the near shore. The study has the potential to be expanded to include the regular capture of survey and monitoring data and the development of a ‘survey spectral library’ of hand gathering across a wide temporal period, including prior and post-introduction of byelaws and management initiatives.

SAMS provided three key outputs for Sussex IFCA:

(a)          a Standard Operating Procedure (SOP) for their drone deployment for HG Monitoring,

(b)          a data and post-processing strategy and approach for data captured by drone, and

(c)          a remote sensing classification analysis (proof-of-concept) using Deep Learning methods.

The results from the study showed that Deep Learning successfully detected hand gathering activity with moderate accuracy within the historical imagery, and further activity with moderate accuracy within the recent imagery.