A systematic map of the technologies used to observe human presence, movement, and activity across parks, green space, and protected areas — and where the methods are heading.
Two databases, screened in stages against the inclusion criteria. Most records fell away at title and abstract; 246 studies form the final corpus.
Publications take off after 2018 and peak at 43 in 2022. Yet computer vision (terracotta) and hybrid (violet) appear only after 2019 and stay a thin band on top — the surge is almost entirely sensor-based.
Of 246 studies, 196 (80%) rely on sensors — overwhelmingly GPS. Only 28 use computer vision and 22 blend both. GPS alone accounts for 183 of 246 studies.
Sensors cover every behavioural dimension, owning movement (70) and presence (122). Vision and hybrid methods cluster in activity & interaction (22, 20) — recognising what people do — but barely touch movement (5, 2). Those faint, dashed cells are the open frontier.
How the corpus flows from sensing hardware, through the algorithm family that interprets it, to the behaviour it ultimately measures. GPS feeds location-based pipelines into all three dimensions; cameras feed computer-vision pipelines almost entirely into activity recognition.
Systematic search · Web of Science + Scopus
246 studies · 2010-2024