Thanks for the answers. We're planning to make a system/prototype, something robotic, for our thesis that can be used after a disaster for search and rescue, and also finding dead bodies. We're planning to use image processing that can detect partial body parts (I just realized that I posted in Arduino section) and several sensors like PIR or the d6t for the project. Those may work for finding humans that are still alive but I don't think using image processing alone is sufficient for detecting a dead body. What about using air quality sensor?
How close do you plan to be to where you're searching?
If you plan to be a fair distance to use cameras/image processing, OpenCV is a Python library that does EXCELLENT at customizing it. There's even a couple of facial recognition apps for it (NOT just facial detection!). You can program it to track things such as a ball, even after half of the ball disappeared behind something. It may be of great use for this project! Also, someone else developed a neural network based on Python libraries, which I never linked even though I intended to.... but if you're familiar with coding, you may be able to!
By sensors alone, it still stands you'd need a sensor at least SOMEWHAT specific to what you're trying to find. A general or generic air quality sensor would go crazy usually after let's say some form of storm in example, where there are gas leaks.... which even may have been the CAUSE of some of the deaths of the theoretical dead bodies you're trying to find.
Finding an airborne chemical a dead body releases, other than a carbon monoxide compound, would probably be most imperative to sorting out the surroundings.
As an afterthought, I just remembered, there ARE some nightvision/IR cameras for the Raspberry Pi, which can run the OpenCV library I was talking about. It would require a fair amount of processing power to run OpenCV for multiple limbs though, so you'd not want anything less than the Raspberry Pi 3, in terms of power. The camera for it, regular or nightvision, can be found for about $30. If you're feeling explorative, there's also ways to turn an existing camera into "nightvision", though I've never done so myself.
That's about as far as I think you're going to go with electronics on the consumer market for the time being.
How close do you plan to be to where you're searching?
If you plan to be a fair distance to use cameras/image processing, OpenCV is a Python library that does EXCELLENT at customizing it. There's even a couple of facial recognition apps for it (NOT just facial detection!). You can program it to track things such as a ball, even after half of the ball disappeared behind something. It may be of great use for this project! Also, someone else developed a neural network based on Python libraries, which I never linked even though I intended to.... but if you're familiar with coding, you may be able to!
By sensors alone, it still stands you'd need a sensor at least SOMEWHAT specific to what you're trying to find. A general or generic air quality sensor would go crazy usually after let's say some form of storm in example, where there are gas leaks.... which even may have been the CAUSE of some of the deaths of the theoretical dead bodies you're trying to find.
Finding an airborne chemical a dead body releases, other than a carbon monoxide compound, would probably be most imperative to sorting out the surroundings.
As an afterthought, I just remembered, there ARE some nightvision/IR cameras for the Raspberry Pi, which can run the OpenCV library I was talking about. It would require a fair amount of processing power to run OpenCV for multiple limbs though, so you'd not want anything less than the Raspberry Pi 3, in terms of power. The camera for it, regular or nightvision, can be found for about $30. If you're feeling explorative, there's also ways to turn an existing camera into "nightvision", though I've never done so myself.
That's about as far as I think you're going to go with electronics on the consumer market for the time being.
Thanks! I just actually started learning OpenCV last week. Last question. If I am about to use a drone that is as small as possible, is it better (if possible) to just do the image processing on a laptop or computer by using the transmitted video from the drone instead of using a camera connected to Raspberry Pi mounted at the drone?
