What Is Camera Trapping? Letting the Animal Take the Photograph
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
A camera left in the forest, triggered by the heat and movement of a passing animal, records what walks past it for weeks without anyone present. That removes the observer from the observation, which is the whole point, since many animals are nocturnal, wary or simply too rare for a person standing in a wood to ever see.
How the trigger works
Modern camera traps are triggered by a passive infrared sensor, which detects a change in infrared radiation across its field of view, meaning it responds to something warmer than the background moving relative to it. That design consumes almost no power while waiting, which is what allows a camera to run for months on batteries. It also produces the technique's characteristic failures: it misses animals whose body temperature is close to ambient, which is why reptiles and amphibians are badly detected, it triggers on vegetation moving in sun and wind, producing thousands of empty images, and it has a detection zone that does not match the camera's field of view, so animals are frequently photographed partly out of frame or missed entirely at the edges. There is also a delay between trigger and exposure, which is why fast-moving animals appear as a tail leaving the picture, and reducing that delay is one of the main things distinguishing research-grade cameras from cheap ones.
What it is used for
The applications extend well beyond photographing animals:
- •Confirming presence, which for extremely rare species is a result in itself and has rediscovered animals thought extinct
- •Estimating density for individually identifiable species, since tigers, leopards, jaguars and others have unique markings that allow a photograph to identify the individual, making capture-recapture analysis possible without capturing anything
- •Measuring activity patterns, since every image carries a timestamp, which reveals whether a species is nocturnal, crepuscular or shifts its timing where people are present
- •Recording behaviour and interactions that nobody would witness, including predation, scavenging and use of trails
- •Monitoring human activity, since cameras set for wildlife record poachers, which has made them an anti-poaching tool and a surveillance concern simultaneously
- •Long-term monitoring at fixed points, which detects change over years in a way that occasional surveys cannot
The data problem
A camera network generates an enormous volume of images, the great majority of which contain nothing, and reviewing them was for years the bottleneck that limited how many cameras a project could deploy. Volunteers reviewing images through online platforms addressed part of it and created a substantial citizen science community. Machine learning has largely solved it: models trained on labelled camera trap images now filter empty frames and classify species with accuracy high enough for routine use, and shared platforms provide those models to projects that could not train their own. That has shifted the constraint from analysis back to deployment and to study design, which is where it should be. Design matters more than most projects allow, since placing cameras on trails because animals use them produces data about trail users rather than about the landscape, and estimating density requires either random placement or an explicit model of how placement relates to movement, which is a statistical question rather than a fieldcraft one.
The ethical questions
Two issues recur. Cameras deployed in inhabited landscapes photograph people, including people who have not consented and who may be doing nothing wrong, and research has documented communities in several countries objecting to cameras placed near settlements and paths without consultation, with the concern being sharper where conservation enforcement and local land use are in conflict. Guidance now generally requires consulting communities, signing camera locations and having a policy for images of people, and several projects delete them unexamined. The second issue concerns publishing locations, since precise coordinates of a rare animal are useful to poachers and collectors, which has led journals and databases to permit withholding location data for sensitive species, a departure from the usual push towards open data that is justified by the specific risk. A smaller question concerns disturbance, since infrared flash is less disruptive than white flash but not nothing, and some species detect and avoid cameras, which biases the data.
The takeaway
A camera trap fires when a passive infrared sensor detects something warmer than the background moving, which uses almost no power while waiting and which explains why reptiles are missed, why wind triggers empty frames and why fast animals appear half out of shot. It confirms presence, estimates density for individually marked species through capture-recapture without capture, and records activity timing. Machine learning removed the image review bottleneck, and cameras in inhabited landscapes photograph people, which requires consent policies.