PIR Sensor ultra sensitive

If they are localised to the one area and slow-moving, then I would start with Shabaz's suggestion of a simple camera with intervalometer.

Initially set it to take an image every minute or so and leave it running for a day. Review the captured images by using a photo viewer app that allows you to quickly flick through a folder of images until you find any activity. Or you can turn the image sequence into a video and watch the video which will then playback at the frame rate you encoded at, typically around 30fps.

If you don't want to tie up an existing camera all day, then perhaps look at a cheap 'nature cam' which usually have intervalometer features.

You can then reposition the camera as necessary and then reduce the time interval to get more detail.

The issue you may have is that they may enter the room elsewhere then quickly fly to the location you are seeing them in before they start crawling slowly.

Comparing file sizes may not work for you here depending on how the light varies, the file compression being used and how much of the image changes. I have successfully used it in the past for detecting the likes of sunrise and sunset though.

You might have more success with the likes of OpenCV for this.

   Images Comparison with Opencv and Python
   https://www.youtube.com/playlist?list=PL6Yc5OUgcoTlQuAdhtnByty15Ea9-cQly

If it is to be a longer term project then perhaps a R Pi with camera running OpenCV to automate the process. 

If they are localised to the one area and slow-moving, then I would start with Shabaz's suggestion of a simple camera with intervalometer.

Initially set it to take an image every minute or so and leave it running for a day. Review the captured images by using a photo viewer app that allows you to quickly flick through a folder of images until you find any activity. Or you can turn the image sequence into a video and watch the video which will then playback at the frame rate you encoded at, typically around 30fps.

If you don't want to tie up an existing camera all day, then perhaps look at a cheap 'nature cam' which usually have intervalometer features.

You can then reposition the camera as necessary and then reduce the time interval to get more detail.

The issue you may have is that they may enter the room elsewhere then quickly fly to the location you are seeing them in before they start crawling slowly.

Comparing file sizes may not work for you here depending on how the light varies, the file compression being used and how much of the image changes. I have successfully used it in the past for detecting the likes of sunrise and sunset though.

You might have more success with the likes of OpenCV for this.

   Images Comparison with Opencv and Python
   https://www.youtube.com/playlist?list=PL6Yc5OUgcoTlQuAdhtnByty15Ea9-cQly

If it is to be a longer term project then perhaps a R Pi with camera running OpenCV to automate the process. 

This is a long term project as I have been at it for 2 seasons now (about a year) I will try these image comparison suggestions and see how it works. Since this will be over 24 hr observation period lighting will vary from day to night. I'm guessing I will have to have bright lights on the area to minimize ambient variations.......

I will post my results after I learn and  experiment with the software solutions. I conclude from all this feedback that a PIR sensor is not practical for this application.

For a longer term bug hunt solution you could perhaps consider adding a pan-tilt hat into the equation. That way you can run multiple interval captures in parallel, potentially covering a larger area.

A camera slider might be another option for this type of surveillance as you can run the camera parallel to a wall.

https://shop.pimoroni.com/products/pan-tilt-hat

https://www.adafruit.com/product/3353

https://www.waveshare.com/pan-tilt-hat.htm

https://www.arducam.com/product/arducam-pan-tilt-platform-for-raspberry-pi-camera-2-dof-bracket-kit-with-digital-servos-and-ptz-control-broad-b0283/

Lighting will indeed affect this, both in colour temperature and intensity. If you are using automatic iris on a camera then your depth of field will also reduce in lower levels of light potentially causing issues as well. If you are comparing between successive images though then it can be less of an issue as less variation between captures. Also some pre-processing can be done before the comparison and thresholds can be adjusted.

You could perhaps attach a LED illuminator to the camera so as you are only illuminating the area being captured and keep it at a consistent level. 

Ultrasonic sensors might be worth a look though as often insects can make a lot of noise at specific frequencies.

beacon_dave said:

Ultrasonic sensors might be worth a look though as often insects can make a lot of noise at specific frequencies.

You may be on to something. I know Ultrasonic transducers/emitters are claimed to be able to repel rodents and insects etc. Maybe it could work in this scenario.

Came across this blog that provides one persons experiences (with ants).

https://www.tomsguide.com/news/i-used-an-ultrasonic-pest-repeller-against-an-ant-invasion-heres-what-happened