Technique to detect flower‐visiting insects in video monitoring and time‐lapse photography data

Abstract

Recent developments in video monitoring and time-lapse photography techniques have increased the efficiency of observing floral visitors. However, such approaches still require a great investment of time and effort to extract the relevant information from the large quantity of data produced. In an attempt to shorten this time-consuming process and make it more efficient, we have developed a system to automatically detect moving objects. This new system involves the comparison of consecutive frames to detect differences between the images and highlights those frames that are likely to contain insect visitors. The specifications of the system and the results from a pilot study using the data of the orchid species Platanthera ussuriensis are described below.

Deep convolutional neural network for automatic discrimination between Fragaria × Ananassa flowers and other similar white wild flowers in fields

“…in the typical flower images [ 6 , 22 , 34 ]. The complex backgrounds also make the discrimination task more difficult, for risking probably discriminating background scenes rather than the object itself [ 18 , 23 ]. Perhaps the greatest challenge originates from the intra-category versus inter-category diversification, i.e.…”

The use of digital video recorders in pollination biology

“…Direct observation techniques have long been considered an effective sampling technique and they continue to be an effective tool in the monitoring of pollinator behaviour (e.g. In recent times, cameras and video recordings have been used in pollination biology (Bumrungsri et al, 2009;Steen & Thorsdatter Orvedal Aase, 2011;Lihoreau et al, 2012;Lortie et al, 2012;Nakase & Suetsugu, 2015;Steen, 2016) with the optimum technique dependent upon the type of flower visitor expected and the study plant morphology. However, this approach is highly labour-intensive and is also limited by observer bias (Westphal et al, 2008;Popic et al, 2013).…”

“…As such they have gained widespread use in marine research documenting trawl net interactions with wildlife (Ferrari et al, 2015), monitoring fish assemblages (Letessier et al, 2013) and even documenting seagrass dispersal (Darnell et al, 2015). Terrestrial uses for the technology are growing, with documentation of prey cue preferences in the dusky pygmy rattlesnake (Sistrurus miliarius barbouri) (Holding et al, 2016), of the feeding behaviour of large mammals (Owen et al, 2015) and some use in pollination biology (Steen & Thorsdatter Orvedal Aase, 2011;Lortie et al, 2012;Edwards et al, 2015;Nakase & Suetsugu, 2015). The successful utilisation of action cameras in such studies demonstrates that they may offer a suitable method for making valid and accurate assessments of pollinator assemblages, abundance or behaviour, with less investment of time, allowing greater replication over a broader scale.…”

“…Digital video recording (DVR) has the potential to allow more efficient observation, with a single researcher able to conduct multiple simultaneous observations over a range of temporal and spatial scales. In recent times, cameras and video recordings have been used in pollination biology (Bumrungsri et al, 2009;Steen & Thorsdatter Orvedal Aase, 2011;Lihoreau et al, 2012;Lortie et al, 2012;Nakase & Suetsugu, 2015;Steen, 2016) with the optimum technique dependent upon the type of flower visitor expected and the study plant morphology. Steen (2016) demonstrated the applicability of automatic camera monitoring of pollinators, utilising a motion detection device in a small-scale (single inflorescence) survey.…”

Insect Visitors and Potential Pollinators ofOrchis militaris(Orchidaceae) in Southern Belgium

“…Indirect approaches have been developed to study more easily orchid-pollinator relationships: pollinaria carried by insects (caught on rewarding plants growing in the vicinity of studied orchid populations) allow the identification, through molecular techniques, of the corresponding orchid species ( Widmer et al 2000 ); pollinaria found on insects can also be identified through specific morphological features ( Singer et al 2008 , Micheneau et al 2009 ); scales and hairs found on pollinated stigmas can allow the identification of the corresponding insect species ( Micheneau et al 2009 ). Videotape observations are another alternative to direct observations ( Micheneau et al 2009 , Lortie et al 2012 , Nakase and Suetsugu 2016 ), but observation-based techniques don’t allow accurate identification of numerous insects (especially dipterans and hymenopterans). The latter have to be collected and examined carefully in order to draw up accurately visitor and potential pollinator lists.…”