I'm an electronics engineering student and I am going to use a sensor that detects infra-red emitted by birds that invade rice paddies.
Do birds emit infra-red radiation?
Birds emit infrared.
Objects with a temperature higher than the background emit detectable infrared (IR). Endothermic (warmblooded) animals keep their body temperatures at around 37oC and given the relatively cool temperatures at the earth's surface, endotherms generally emit more IR than the background. Endothermic animals include the mammals and birds, but also some fish.
Infrared image of a hummingbird. Source: Nature Conservancy.
Also note your IR recordings will benefit from lower temperatures and reduced light-pollution from the sun, hence your recordings will be better at night than during daytime.
- Grady et al., Science (2014); 344(6189): 1268-72
Dinosaurs is a very broad term which includes both the ancestors of birds as well as modern reptiles. But that analogy stretches as far to say that a bird is a modern dinosaur and a reptile is a modern dinosaur but a bird is not a reptile. Both of their ancestors lived during the cretaceous period and tend to get lumped together.
Another analogy would be that dolphins, bats, and humans are all modern-day mammals. But a dolphin does not fly and humans do not live underwater, etc.
Modern-day birds descended from a group of dinosaurs called theropods which were warm-blooded dinosaurs.
See here for more information:
To be specific to your question, if it is an endotherm then it will emit some level of infrared radiation, as this is their primary means of avoiding overheating.
Interestingly, some birds are actually warmer than humans at 102-112F:
Every object hotter than 0K emits infrared radiation. Many object absorb and reflect, re emission infrared radiation (usually lower frequency). For detect bird need some difference with background object's infrared radiation, like different intensity, spectral, phase. One person above show use intensity (at night ground/trees become cool emit less radiation than birds). If need find birds during day, and ground hot, birds appear darker than ground.
Can also use multiple infrared sensors together detect different frequency (like visible light, infrared light have frequency range) different object have different infrared color. Also, if rice paddies have water, reflected infrared light is polarized but reflection from bird is random, maybe can use this help find bird when they fly. I think sky UV and visible light also is polarized (bee use it for navigation), I guess infrared light from sky is polarized too.
Typically an animal must be 2.7°C warmer than its surrounding environment, to trigger a detection.
Are you talking about herons that hunt frogs or small birds that forage seeds?
The distance of an animal from the camera’s PIR sensor and its body mass, varies the magnitude of infrared radiation, which affects the ability of the PIR to trigger an image. So PIR triggers perform well for large and medium sized mammals, but do not reliably detect ectotherms (reptiles, amphibians, and invertebrates) or smaller mammals.
I'm drinking a beer while i write this so I can answer well, and by unabashed plagiarism: here is a complete review of some PIR technologies, which is biased towards their 3mm laser sensor called HALT: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5628828/
Personally i'd say forget about complex electronics for this project! just take a slow exposure image every seconds and pass the image stack through pixel processing code to search for bird sized dots and lines.
Birds in paddies should be easy to detect, because they come down from the sky which has a light background and they move at relatively high speed. So a computer can easily recognize the trails against the sky and classify them by size. You can therefore use a very cheap time lapse camera trap which takes 1000ds of pictures a day, at a cost of perhaps 1 gygabites per day, and do comprehensive statistics on bird movement around the paddy fields. Infra red? home made? i'd say forget about it, image recognition digital signal processing is more thorough.
here are some pics from here
Recently I programmed image stacks, and it was very much fun because it was 2d and 3d graphs. to sequence through an image file is 12 lines in U3D. To loop through the pixels is another 12 lines, so all you have to do is write conditions. to save relevant images to another area. That said, you have 1 minute exposure images all day, you will have 600 images a day to sort through manually.