Accurate estimation of fish abundance and demographic structure is critical for assessing ecosystem condition and informing conservation management. Non-invasive surveys, particularly remote video, are increasingly used in ecological surveying; however, existing analyses are unable to distinguish individuals (of the same species) within video files, unless they are observed at the exact same time or video frame (Williams et al. 2022; Harvey et al. 2001). This can lead to underestimation of true abundance and life-stage composition when individuals are not observed simultaneously.
Here, we present MIDS (Multi-frame Individual Distinctions by Size), a novel analytical framework that uses length measurements, precision error, and statistical inference to distinguish individuals across multiple frames within a video sequence, regardless of when the individual(s) appear(s). We applied MIDS to stereo-video deployments across two estuaries in New South Wales, Australia, and compared outputs with the widely used single frame ‘MaxN’ metric. In comparison, MIDS consistently detected higher abundances across all camera deployments and improved resolution of life-history stages. The largest increases were found within juvenile cohorts, which included an additional one-hundred and ten (110) juvenile Yellowfin Bream (Acanthopagrus australis) in one estuary.
By improving abundance and size-structure estimates from non-invasive surveys, MIDS provides a robust framework for ecological monitoring and management, enhancing the reliability of data used to inform conservation policy and resource management decisions. Importantly, the utility of MIDS extends beyond its current application; By enabling the identification and removal of repeat observations across time, MIDS improves data extractability with multiple spatially separated cameras, longer sampling durations, and emerging AI-driven workflows for automated length estimation and species identification. Collectively, these advances position MIDS as a scalable and future-ready tool for improving ecological inference in aquatic systems.