Vape Detection / Alarm

[psynegy]

Has anyone found a (relatively) cheap option for vape detection in toilets?

 

Verkada dectectors are mad (and ongoing) money, all the other seem to be US companies without pricing on their websites.

 

Considering knocking up some myself with some IoT arduinos and PM2.5 sensors.

 

Hoping I can save myself a job! Thanks!

[Aprice]

Also thought about giving this ago, ended up buying some SGP30 sensors from Pimironi and got them working quite quickly. I'm just not quite sure how you'd go about identifying vapes over other substances. I'm sure the vapes must be full of VOCs, but not sure what else would trigger it.

 

Maybe PM2.5 is a better way of doing it, the problem we have is that the students are already spraying deodorant into the smoke detectors in the toilet and I'm not sure how well they'd hold up to it.

 

Did think about putting the sensor into the extraction vent in each cubicle, I thought that might give it a better chance of detecting and protect it from vandalism a bit?

Edited January 19, 2023 by Aprice

[synaesthesia]

Verkada are Raspberry Pi based but VERY much worth the money. The SV11 are on sale at the moment - they're not as expensive as you think. It's pretty much the software you're paying for, but the hardware is very well built and runs on PoE.

Halo is the other name worth looking at. Do not get sucked in by fly by nights like SMF Systems who are flogging chinese whiteboxes worth £18 for 10 times their worth, using consumer grade "smart device" software to manage.

Verkada at least has PM2.5 and TVOC separate from vape detection so false positives are mitigated, plus with motion and noise detection it makes it worth while.

 

Addition:

 

Hopefully this helps show the difference a little - here's the difference between the PM2.5 detection and specific Vape detection for the same sensors, so you can see the distinct correlation. You'll be able to do it with just PM2.5 but expect the vape is a mixture of that and the air quality index. TVOC doesn't flag for vape.

 

This is the Vape index:

 

And the PM2.5:

 

And air quality:

 

Looking at pm2.5 alone, you will be getting constant false positives from female scents - perfumes far more than male deodorant trigger this highly.

Edited January 19, 2023 by synaesthesia

[howartp]

We've got four Verkada SV11's with Verkada cameras and about to add another four.

 

SLT love catching kids, but then moan that the timing of the alert doesn't match the time on the camera - they've tested with member of staff that vapes, but i've never been around when they've done it.

 

I reckon students are vaping in one cubicle at the far end from the sensor, and it's taking time for the vape to reach the sensor at the other end.

[synaesthesia]

Indeed, there's a small scope for this to work reliably - so if there's more than 2 cubicles, you'll need another sensor.

Make sure they're synced with the same time server - we don't have any issues with timing.

[dhicks]

Also thought about giving this ago, ended up buying some SGP30 sensors from Pimironi and got them working quite quickly.

 

I've looked at them, thinking for general CO2 / air quality monitoring. Do you have multiple Raspberry Pis around the building collecting data? If so, how do you gather that data into some kind of report and/or get alerts when there's an issue?

[TechMonkey]

Not as expensive as you think for Verkada must be relative. We were quoted around £510 a sensor, which is not something we can afford, especially when there are 12 toilets that would need kitting out.

SV11 has been usurped, so they are selling them cheap to make way for the new sensor (that is my interpretation so may be me being cynical though). Be careful as the rep I spoke to was pushing the SV21 and I had to point out that it does not do vape detection.

 

If you have floor to ceiling cubicles you need a sensor per cubicle. Kerching.

 

I'd love to sensor everything up. I've seen some that can pick up aggression, naughty words, estimate numbers and loud noises, which would all be great to detect. But the cost is a concern, and the license is another annual payout on the budget.

[dhicks]

Verkada are Raspberry Pi based but VERY much worth the money.

 

Ah - interesting. We had one of their salespeople in to talk about their door-entry devices, I didn't speak to them but I gather we weren't impressed by them, however your actual implemented experience sounds very good. I gathered their CCTV cameras were re-badged Hikvision devices, managed by their own software (although possibly still with Hikvision firmware), would you have any idea if that is correct? Am I right in thinking that when you say "Raspberry Pi based" you are talking about their environment monitoring kit, or are other bits of equipment (door entry, CCTV cameras) based on Raspberry Pis too?

[synaesthesia]

Unsure on the cameras, we've only got 6 sensors in (SV11) working alongside our existing CCTV. I haven't had the SV11's apart but they picked up on DHCP and their mac addresses show they are RPi3 based.

 

I can't go into costing specifics but ours were WAY less than £510 a unit. Worth asking around, there's quite a few resellers of the gear in the UK.

[kearton]

I can't go into costing specifics but ours were WAY less than £510 a unit

#intrigued

 

We've got four Verkada SV11's with Verkada cameras and about to add another four.

Can you reply (or PM me) with details of where you got them from and how much they cost please?

[dmj]

I can't go into costing specifics but ours were WAY less than £510 a unit. Worth asking around, there's quite a few resellers of the gear in the UK.

My guess would be an ESP32, a VOC sensor, a printed case and a some software which is all of about £15 in components

[synaesthesia]

My guess would be an ESP32, a VOC sensor, a printed case and a some software which is all of about £15 in components

 

For the verkadas? Not a chance. As said above, it's way more than just a VOC sensor, because that alone isn't sufficient for reliable vape detection.

[dmj]

For the verkadas? Not a chance. As said above, it's way more than just a VOC sensor, because that alone isn't sufficient for reliable vape detection.

 

hmm, interesting. I wonder if they've got a specific sensor, maybe for ethylene glycol.

[synaesthesia]

As I put in my original post, I imagine it will be a combination of the sensors and the software doing the grunt. I wouldn't mind knowing what's in the Halo sensors for their prices. We tested a chinesium device based on an "Air Bath®" with a senseair.com detector and "WBR3" combi bluetooth & wifi. Parts that literally cost £8 and the "UK Company" shifting them are charging 10 times that. They're in generic plastic boxes which would otherwise house ethernet switches, routers, usb hubs and connect with those awful home smart apps.

It's not always technical too - you need to think legally. Let's say you exclude a student where part of the evidence is as a result of one of these devices. With 2 of the big names, you'll have support should you need it in a review hearing. Will not be that easy otherwise, it could be made entirely inadmissable.

[dmj]

As I put in my original post, I imagine it will be a combination of the sensors and the software doing the grunt. I wouldn't mind knowing what's in the Halo sensors for their prices. We tested a chinesium device based on an "Air Bath®" with a senseair.com detector and "WBR3" combi bluetooth & wifi. Parts that literally cost £8 and the "UK Company" shifting them are charging 10 times that. They're in generic plastic boxes which would otherwise house ethernet switches, routers, usb hubs and connect with those awful home smart apps.

It's not always technical too - you need to think legally. Let's say you exclude a student where part of the evidence is as a result of one of these devices. With 2 of the big names, you'll have support should you need it in a review hearing. Will not be that easy otherwise, it could be made entirely inadmissable.

 

Yeah it makes sense that it is a combination of things. I couldn't find a specific ethylene glycol electrochemical sensor, I guess the chemistry is too similar to other VOC's. The ultimate would be a double focusing mass spectrometer, but that would be a bit too expensive! Perhaps an infrared spectrum would work, that's sometimes used for perimeter analysis to warn of large scale chemical leaks again probably too pricey.

 

It would be interesting to know how perfumes are detected on the verkadas.

 

FWIW I'm not convinced by the inadmissible argument - the HT would search the student anyway to look for contraband and a sensor is really to inform the decision to search.

[howartp]

Our mac addresses are all E0:A7:00 which is assigned to Verkada; I'm not familiar with Pi's so can't comment.

 

Can you reply (or PM me) with details of where you got them from and how much they cost please?

We buy from RV Fire Systems in Lancashire. They do all our fire alarm support/maintenance but the vape stuff is supply-only for our sparkies to install and me to commission.

 

I can't go into costing specifics but ours were WAY less than £510 a unit.

I've just looked at an invoice. Cough, ahem... ours are NOT way less than £510 per unit! I'm not involved in procuring but...!

[synaesthesia]

Depends when you bought/ordered them - I think we got lucky as they were recent orders, just prior to being announced EOL. Either that or our trust procurement team are miracle workers.

 

And yeah, our MAC's are the same. They are reporting twin MAC addresses though at points on our dhcp server. It may also be a larger combination of hardware. The other mac's pick up as B8:27:EB.

 

They're certainly intriguing devices and keen to know more about how they work - compared to the chinesium devices we have, there appears to be a lot going on.

[dhicks]

Do you have multiple Raspberry Pis around the building collecting data? If so, how do you gather that data into some kind of report and/or get alerts when there's an issue?

 

Just looking a bit further at this today. Some very rough pricing:

 

Pimoroni Raspberry Pi Zero WH Essentials Kit: £42

Pimoroni Breakout Garden Mini: £10.80

Pimoroni SCD41 CO2 Sensor Breakout: £54.60

 

That's £107.40 for parts, not including an exact-sized case to hold everything. So far I've found adafruit.io, freeboard.io and thingspeak.com that have some kind of IoT-type data storage facility, although I'm not really sure how any of those work quite yet.

[Aprice]

Looking at the data sheet It has a VOC sensor, Particulate matter sensor, Motion, Temp and Humidity sensor and Noise.

 

Also has some very handy specs of the sensors it's using.

 

I wouldn't have thought it would be all that difficult with that set of sensors to get a good idea of vaping.

 

Just need to gather some data!

[Aprice]

I've looked at them, thinking for general CO2 / air quality monitoring. Do you have multiple Raspberry Pis around the building collecting data? If so, how do you gather that data into some kind of report and/or get alerts when there's an issue?

At the moment just connected to an Arduino and printed the sensor output out the serial port, just to have a play. The idea was to develop something like the Verkada, just never got any further. Now it seems there's some interest from others I might have another go at it.

Edited January 20, 2023 by Aprice

[dmj]

Just looking a bit further at this today. Some very rough pricing:

 

Pimoroni Raspberry Pi Zero WH Essentials Kit: £42

Pimoroni Breakout Garden Mini: £10.80

Pimoroni SCD41 CO2 Sensor Breakout: £54.60

 

That's £107.40 for parts, not including an exact-sized case to hold everything. So far I've found adafruit.io, freeboard.io and thingspeak.com that have some kind of IoT-type data storage facility, although I'm not really sure how any of those work quite yet.

 

Raspberry pi is probably a bit OTT for this sort of thing (plus they are quite hard to get hold of at the moment). It should be more than possible to use an ESP32/ESP8266/Arduino for this, send the results to an MQTT broker running onm a server and then publish/process from there.

[dhicks]

Raspberry pi is probably a bit OTT for this sort of thing (plus they are quite hard to get hold of at the moment).

 

Indeed, although we have a tub of Pi 3s and 4s spare at the moment, so I plan to use a Pi to get a basic temperature sensor working in the server cupboard and see where we go after that.

 

send the results to an MQTT broker running onm a server and then publish/process from there.

 

Ah ha - what would you recommend? I was looking at adafruit.io, but even the paid-for version (60 data points per minute) seems like it might not scale if we wind up with (say) each classroom having a sensor and reporting half-a-dozen measurements every few minutes. I could, of course, write something basic to listen out at an HTTP(s) endpoint to collect data (devices would do "post apikey=123456 devicekey=123456 tempc=29") and have it dump that data somewhere periodocally, but I figured there must already be a tested / supported solution for that. RabbitMQ seems like it might be a solution, but I'd still need to write something that runs as a process that collects and processes the queued data periodically.

[dmj]

Ah ha - what would you recommend? I was looking at adafruit.io, but even the paid-for version (60 data points per minute) seems like it might not scale if we wind up with (say) each classroom having a sensor and reporting half-a-dozen measurements every few minutes. I could, of course, write something basic to listen out at an HTTP(s) endpoint to collect data (devices would do "post apikey=123456 devicekey=123456 tempc=29") and have it dump that data somewhere periodocally, but I figured there must already be a tested / supported solution for that. RabbitMQ seems like it might be a solution, but I'd still need to write something that runs as a process that collects and processes the queued data periodically.

 

https://mosquitto.org/ is a widely used MQTT broker for self hosting which would scale. IoT core is the Amazon cloud equivalent: https://aws.amazon.com/iot-core/

I recently followed this energy monitoring project, and it has examples using IoT core. The code is available and should be fairly easy to add/adapt functionality for VoC/CO2 monitoring.

https://savjee.be/blog/home-energy-monitor-v2/

https://github.com/Savjee/home-energy-monitor

[Disease]

We have just Bought 3 x Verkada SV23 after having one on trial for 2 months, works really well and the detection range is really good, caught a number of Students vaping.

[psynegy]

We have just Bought 3 x Verkada SV23 after having one on trial for 2 months, works really well and the detection range is really good, caught a number of Students vaping.

Out of interest, who was the trial unit arranged with?

 

Edit: also someone mentioned the new SV23's don't do vape detection? Is that not the case?

 

Edit edit: apparently that was the SV21's...! Ignore me!

Edited January 23, 2023 by psynegy