I posted earlier about putting out my mason bee houses. I haven't been very successful with bees nesting the last couple of years so I thought this year that I would put up a camera to monitor the activity at the bee house. I did a couple of small AI camera projects on Hackster last year so my first thought was that I would try to use one of those to detect bee activity. The first project was the Dual AI Camera that used a Seeed Grove Vision AI V2 and Xiao ESP32S3 Sense for hummingbird detection and the second was the RPi Zero W Person Sensor Testbed which used an RPi AI Camera with an RPi Zero 2 W for Person Sensor testing. In either case it would require training and deploying an appropriate model for bee detection. The other major issue is that it is still raining here quite a lot and neither of these projects is weatherproof so I would need to build waterproof housings.
In the short term I decided that I would deploy a TP-Link Tapo C121 Security Camera that I have. This is a nice compact and inexpensive 2K IP camera and it is IP66 rated. It has motion detection and free Smart AI Detection for people, pets, and vehicles. It also allows for RTSP streaming and local SD storage so that I won't need to access the cloud for recording. I won't be able to do bee detection (unless I try to deploy a model on Frigate) but for now I'll try to see how it goes just trying to use motion detection. Hopefully longer term I'll be able to deploy a solar powered dedicated AI camera with a mason bee detection model.
I mounted the camera on a metal garden pole that I used in a previous challenge. The camera has a magnetic baseplate so it snapped right onto the top flange on the pole. I tried not to put the camera too close to the bee house so I wouldn't disturb the bees. I also turned off the visible and IR LEDs and the status LED.
Unfortunately, I haven't been able to identify mason bees in either the front or back yard yet. I think when it gets sunnier I'll have more opportunities to see them at rest. The closest I've gotten so far is that I saw a native bee on a leaf in the front yard but because its wings weren't deployed I couldn't get a clear view of the body to identify it. Definitely a bee because of the distinct antennae.
After looking at a couple of detection captures I realized that it may be difficult to reliably identify mason bees vs other insects with the resolution and distance that I'm using. Here's one where I drew a box around a bug on the right edge of the house. Not enough resolution to get a clear image even if I zoom in.
I did see something I thought was a bee on the camera and went out to take some pics with my iPhone.
But with a zoomed in view it is clearly a housefly.
I tried limiting the motion detection zones but the camera is still detecting activity (mostly squirrels) along the top of the fence.
The captured video does not contain the bounding boxes so I need to do a screencap to show them.
I was dismayed to see that a bird showed up looking for a meal. I wonder if he's gotten any bees. I guess I'm going to have to put up some sort of screen...
Massive loss of bee colonies in Oregon
Local news stations are reporting large losses in honey bee hives (60%-75%) in Oregon in 2025. Quite worrisome for agriculture in the state.
I am mostly focused on native bees currently but I do watch a community honey bee hive that the dogs and I encounter on our walks. That hive wintered very well. It had been downsized to just the brood box and a feeder. In mid-spring I saw that they had added a super and last week there was lots of activity. Thankfully that hive is doing well. It bodes well for the nearby wild blackberries this summer.
