Photo or Video? Which is Best for Recording Mammals Using Camera Traps?
Blog by Sian Green
Camera traps, or remotely activated cameras, have seen a great rise in popularity over recent years, both with researchers formally assessing mammal populations and enthusiasts eager to monitor their local wildlife. One of the advantages of using camera traps is their ability to record shy and nocturnal species that are rarely seen face to face. This makes them an excellent tool for monitoring many mammal species, as well as providing photos and videos that can be a great resource for organisations engaging with the public. However, sorting through the footage has become a significant and time-consuming challenge for some projects. One solution is to recruit willing volunteers as ‘citizen scientists’.
Data quality has been raised as a concern with citizen science projects, but studies have shown that citizen scientists are able to classify species in photos from camera traps with a high level of accuracy (Swanson et al. 2016; Hsing et al. 2018).
Many models of camera trap can be programmed to take either a series of photos in rapid succession or a video when triggered, yet most citizen science camera trap projects (and indeed most camera trap projects generally) use only photos. There are several reasons why researchers may be wary of using video. The first is a concern that cameras set to video can be slower to react when triggered, and then slower to recover before they can be triggered again, increasing the risk that animals – particularly fast-moving ones – will be missed. Other concerns are that videos use up batteries more quickly and the larger file sizes take up more storage space. They also take longer to process: the videos themselves take time to watch but also, much of the software designed for streamlining the classification process does not yet support video.
Our study
While there are concerns over video, it has many potential advantages. These include behavioural observations, easier identification of species, and increased public engagement with moving images. My supervisors and I work with a lot of camera trap data, and we wanted to understand whether setting a camera trap to video truly had any advantages or disadvantages compared to photo, both in terms of ecological recording and citizen science classification accuracy. We decided to do a formal comparison between these types of data with a paired camera trap study with two cameras fixed side by side at a survey point. One taking photos and the other videos, but with all other settings exactly the same. Our parallel photo and video data sets were then classified both by experts and by citizen scientists via the MammalWeb platform and compared.
What we found
During this survey we recorded 13 types of mammals and 18 types of bird. We chose to focus on eight of the most commonly occurring mammal species (badger, fox, grey squirrel, rabbit, roe deer, muntjac, fallow deer, and wild boar) and for each species used the photo and video data to look at three ecological measures. These included detection,probability the site was ‘occupied’ by an animal and proportion of time spent active by the animal. For all three measures, we found no difference between the photo and video data sets for any of the species showing that setting your camera trap to photo or video can provide the same quality of ecological data.
However, when comparing citizen scientist classifications to expert classifications there were some interesting differences. Overall, citizen scientist accuracy (based on agreement with expert classifications) was higher for videos (95% accuracy) compared to photos (86% accuracy). It is likely that being able to see movement in the video clips makes animals easier to spot and therefore to identify. It was not only accuracy of species identification that differed between photo and video citizen science classifications; people were also more likely to supply age category (adult or juvenile) and sex (male or female) classifications for videos.
Overall I hope our results will encourage more people to consider setting their camera traps to video, especially if they want to engage with citizen science!
If you are interested in reading more, this research is published in an open access paper here: https://doi.org/10.1002/rse2.309
Sian Green is currently working towards a PhD at the University of Durham and is also working with MammalWeb.
The theme for National Mammal Week in 2022 is Recording Mammals Around You. Did you know that mammals are some of the most under-recorded species in Britain! With one in four of our native mammals threatened with extinction, and many others are in decline we need your help. Consider becoming a member of the Mammal Society today for as little as £3 a month.
References
Green, S.E., Stephens, P.A., Whittingham, M.J. and Hill, R.A. (2022) ‘Camera trapping with photos and videos: implications for ecology and citizen science’, Remote Sensing in Ecology and Conservation, doi:10.1002/rse2.309
Hsing, P.-Y., Bradley, S., Kent, V.T., Hill, R.A., Smith, G.C., Whittingham, M.J., Cokill, J., Crawley, D., Stephens, P.A. and Stephens, P.A. (2018) ‘Economical crowdsourcing for camera trap image classification’, Remote Sensing in Ecology and Conservation, 4, pp. 361–374. doi:10.1002/rse2.84
Swanson, A., Kosmala, M., Lintott, C. and Packer, C. (2016) ‘A generalized approach for producing, quantifying, and validating citizen science data from wildlife images’, Conservation Biology, 30(3), pp. 520–531. doi:10.1111/cobi.12695.