picame_vr
Posts: 4
Joined: Thu Jul 17, 2014 8:25 am

Re: Volunteers to assist with this project: Honey Bee Counte

Tue Dec 27, 2016 10:02 pm

Just a quick idea. I've skimmed the thread and the first thought that came to mind was: why not put some kind of white board in front of the hive (if looks are not that important) where the camera is pointing. From what I see the area this board would have to cover is not that great so it would not impact the looks of the hive that much.
I guess this would make it much easier for the algorithms to spot the bees.
Regards,
Picame

P.S: If the white board is to 'ugly', I think even a (dark) grey board or similar would do.

gtoal
Posts: 94
Joined: Sun Nov 18, 2012 12:02 am

Re: Volunteers to assist with this project: Honey Bee Counte

Wed Dec 28, 2016 12:18 am

picame_vr wrote:Just a quick idea. I've skimmed the thread and the first thought that came to mind was: why not put some kind of white board in front of the hive (if looks are not that important) where the camera is pointing. From what I see the area this board would have to cover is not that great so it would not impact the looks of the hive that much.
I guess this would make it much easier for the algorithms to spot the bees.
Regards,
Picame

P.S: If the white board is too 'ugly', I think even a (dark) grey board or similar would do.
Hives of the type our project uses have a landing board in front of the entrance and indeed ours is painted white. Actually any single colour except yellow is probably as good. I expect we'll do colour blob detection on yellow using OpenCV. It may be that a matte black makes exposure control easier but we'll just stay with what we've got. By the way since we first posted we have converted our backyard project into a proper University research project and have moved our hive to the roof of our University library, where we will have a much better network connection that will support better live streaming. I think we'll be back online in February or March. (We're currently waiting for our IT guys to run ethernet and power up to the roof for us, and then there are some bee husbandry things to take care of before we get back into the IoT monitoring side of things.)

Graham

picame_vr
Posts: 4
Joined: Thu Jul 17, 2014 8:25 am

Re: Volunteers to assist with this project: Honey Bee Counte

Thu Dec 29, 2016 1:04 am

gtoal wrote: Hives of the type our project uses have a landing board in front of the entrance and indeed ours is painted white. Actually any single colour except yellow is probably as good. I expect we'll do colour blob detection on yellow using OpenCV. It may be that a matte black makes exposure control easier but we'll just stay with what we've got. By the way since we first posted we have converted our backyard project into a proper University research project and have moved our hive to the roof of our University library, where we will have a much better network connection that will support better live streaming. I think we'll be back online in February or March. (We're currently waiting for our IT guys to run ethernet and power up to the roof for us, and then there are some bee husbandry things to take care of before we get back into the IoT monitoring side of things.)

Graham
Ah, on the roof the floor beneath the hives will be more uniform in color I guess so the detection should improve even more. And fingers crossed that the ethernet runs are installed soon!

gtoal
Posts: 94
Joined: Sun Nov 18, 2012 12:02 am

Re: Volunteers to assist with this project: Honey Bee Counte

Sat Feb 25, 2017 11:50 pm

_electric-hive-small.png
Raspberry Pi in outdoor enclosure
_electric-hive-small.png (59.73 KiB) Viewed 3631 times
hives-on-roof.png
RasPi-Monitored Beehives on the roof of UTRGV Gym
hives-on-roof.png (54.28 KiB) Viewed 3632 times
Sorry for the tiny picture - it appears to be the largest we can upload (64K limit).

These are our hives on the University roof. One of them has a raspberry pi monitoring the entrance, which has a gateway attached that forces the bees to walk in, which slows them down enough for decent image recognition. (This doesn't disturb the bees at all, they adapt to it very quickly)

The Pi also monitors air pressure and humidity, and there is a temperature sensor inside the hive as well as in the pi's enclosure (and another one attached to the PiDrive just in case - we can shut down if the weather is too hot for the drive)

gtoal
Posts: 94
Joined: Sun Nov 18, 2012 12:02 am

Re: Volunteers to assist with this project: Honey Bee Counte

Fri Mar 10, 2017 11:34 pm

Dr DongChul Kim, a new member of our team, is making some progress on using a neural net to classify bee videos: https://www.youtube.com/watch?v=xF09bFD7sS4 - still early days, but I thought you might be interested in seeing the work progress.

deynight
Posts: 1
Joined: Thu Jul 27, 2017 9:27 am

Re: STICKY: Volunteers to assist with project: Honey Bee Counter

Tue Aug 01, 2017 12:56 pm

Hi!
I'm working on a fly counter, same goals but not same insect! I made a system with IR LED/phototransistor, QTR8 sensor from polulu, https://www.pololu.com/product/961 . The mechanism is really similar to this project : http://www.instructables.com/id/Honey-Bee-Counter-II/
It is compose of 2 set of 4 QTR8 sensor so 36 IR LED/phototransistor. The difference is from the sensor I use.

It's actually in test but for the moment the result are not really useful... If you want more information about, let me now and I will explain it.
I want to test other technology and your approach is really interesting, I'd test the program made by Guyver https://github.com/guyver2/beehive/blob/master/bees.py and the result is really encouraging. Some fly were not detect but it seems to be the fps which are to low (30fps...).
So, I have some interrogation about it :
- Some of you (elriri35 , Guyver) have test this program in realtime, did the result were interesting? My objective is to use this system in realtime too so did you have a repo where I can look at your prog ?
- I want to reduce the time of the process, I have convert my video in 640x480 and in grayscale before the process. Is there something else that I can do?
- Here is a screenshot of the video where you can see the fly and the entrance of the box. I've made some change on the parameters from OpenCV but I'm not an expert. Did you have some advice?
frame_00030.jpg
Picture of fly
frame_00030.jpg (29.32 KiB) Viewed 3129 times
For now I have just change the "inertia ratio".

Thanks a lot and I hope that someone is stil on the project :D

TheWaterbug
Posts: 73
Joined: Fri Feb 05, 2016 12:35 am
Location: Los Angeles, CA, USA, Earth

Re: STICKY: Volunteers to assist with project: Honey Bee Counter

Mon Feb 12, 2018 10:39 pm

This looks like a great project! I have zero technical expertise to contribute, but I do have bees :D

I have a live BeeCam here:

https://www.youtube.com/user/IAmTheWaterbug/live

There's nothing Pi about that feed except that it's passing through a Pi working as an ffmpeg relay.

My second camera is built around a Pi:

https://www.youtube.com/channel/UCIVY11 ... RhiMg/live

And that's watching the entrance of a swarm trap. Once a swarm has moved in I could easily reposition the camera to look down on the entrance and place a piece of white plastic under the front porch, to increase the contrast. I could also swap out the standard camera for an IR one.

Or does everyone already have enough data to work with?

Return to “Camera board”