Overengineering a way to know if people are in my university's CS lab

https://www.amoses.dev/blog/upl-people-counter/

by nicosalm • 2 years ago

189 82 2 years ago

amoses.dev

readthenotes12• 2 years ago

Scotch tape. Those poor students couldn't afford duct tape!

ethbr12• 2 years ago

These are CS students, not MEs.

Which is to say, heathens who have yet to accept our lord and savior Duct into their hearts. (And its latter day saint WD-40)

readthenotes12• 2 years ago

I thought wd-40 had been cast down in favor of silicon spray?

fullstackchris• 2 years ago

Love the motivation! Reminds me of the first stuff I built back in my university days.

One thing from and old cranky dev that I notice: it would seem the yml you post has redundancies: either have 4 endpoints, no payload or 4 different payloads, one endpoint (the endpoint itself can tell you what you need to do) However, from the express script it looks like you arrived at this in the final solution anyway. Not sure if I missed something though, is there a reason the API needs such a shape? Cheers!

amosesdev3• 2 years ago

Hi! Author here. The yml is from a configuration in Home Assistant. Whenever it received an event from the sensors, it would fire the corresponding POST request (which you have to specify ahead of time, hence the configuration being a little redundant). The main reason for switching to the Express setup is that I was having a few random people look at the code & send their own requests to fudge what the status appeared as. I could have added some security/authentication to the endpoints, but I liked the idea of having the frontends simply querying the site (or my proxy) instead of having to cache a POST request.

grahamj3• 2 years ago

You can read and write HA entities via HTTP API so I think you could have done this with nothing but HA, but either way it's a fun project :)

One of my more fun HA project was scraping my snow removal service's tractor tracking API and passing the result for the closest tractor into HA so I can be notified (by voice with Alexa and/or push notification) when there's one nearby in case I need to move the car.

alkh2• 2 years ago

I swear to God that all of these CS labs at different unis look the same. I am getting flashbacks of labs in Toronto that looked exactly like pictures in the post

whimsicalism5• 2 years ago

even the physics labs i worked in looked like this

chaboud• 2 years ago

The physics computer lab in Chamberlin Hall at UW in the 90's was a secret treasure trove of idle NeXTstation Turbo machines in an almost always empty room cooled to near refrigeration temperatures. I used to light up at least half of that room to run distributed simulations. There's probably still a 30 year old key to that lab in a junk drawer somewhere.

Eventually I realized that it just made sense to suck it up and get my own hardware, as it was either going to be esoteric "workstation" hardware with a fifth of the horsepower of a Pentium 75 or it was going to be in a room like the UPL jammed with CRT's and the smell of warm Josta.

How do students operate these days? Unless one is interacting with hardware, I'd be very tempted to stay in "fits on a laptop" space or slide to "screw it, cloud instances" scale. Anyone with contact in the last 5 years have a sense of how labs are being used now?

kredd• 2 years ago

That's exactly what I thought! "Is that UofT?"

analog31• 2 years ago

2nd law at work. Just like all cups filled with ice cubes look different, but all cups half-filled with liquid water look the same. ;-)

iwontberude• 2 years ago

I have a stealth startup that designs privacy first solutions to share coarse data about what is inside of rooms without disclosing secrets. This is useful for many use cases, from detecting the number of people in a room to other motion sensing devices which are critical for health and human safety. Patents are pending.

bobbiechen• 2 years ago

That's great! It does feel like cameras generally collect more data than necessary for some purposes. I have a friend who works at Butlr, which uses thermal sensors to detect body heat (and avoid the privacy risks of cameras) - sounds like your startup is in a similar space.

I know you mentioned patents pending, but is there anything you can share about your approach?

Animats• 2 years ago

"Room occupancy sensors" are a common product. Butlr apparently uses a low-rez IR camera, although they avoid calling it that. Passive infrared is common. Millimeter wave is available.[1]

[1] https://www.alibaba.com/product-detail/IoT-Commercial-People...

moralestapia4• 2 years ago

Nice project but it could be illegal. Check your jurisdiction.

Source: have done similar hobby projects for fun, which turned out to be illegal.

soggybread• 2 years ago

What part of this would be illegal? It's just a zigbee door sensor. The only issue I could see is the college getting upset but if anything they'd just say 'take it down'

whimsicalism5• 2 years ago

object detection of people from images is probably what they are referring to. there are maybe some state laws that could be stretched to include that but i would say presumptively legal

whimsicalism5• 2 years ago

this isn't facial recognition

moralestapia4• 2 years ago

???

Who's talking about that?

alexchantavy• 2 years ago

Bunch of restaurants use those fancy Nest cams that remember faces. Is this illegal too? Feels like it should be controlled in some way.

moralestapia4• 2 years ago

Feels like it should be controlled in some way.

It is. Check your jurisdiction.

rithikjainNd01• 2 years ago

UW Madison UPL on Hacker News! Great to see, you guys are awesome!

nicosalm2• 2 years ago

Thank you!

djsavvy• 2 years ago

Neat to see discord channel names being used as an information channel.

thrdbndndn• 2 years ago

It's a common practice/trick in Discord communities (particularly gaming) to pin some status, but yeah, I always find it neat.

teddarific• 2 years ago

This is neat. I worked on something similar to college where we built an app to track how long the line was at a popular cafe. We set up a camera + raspberry pi in the cafe and tried to the number of people. There was a lot of noise from overlapping people, random people walking in and out of line, etc. Cool seeing all the techniques and approaches you tried!

nullhole2• 2 years ago

It then set the name of a channel to the results (either 1-person-in-upl or X-people-in-upl), which others could check.

I'm not suggesting you don't do this, but you /could/ setup a speaker to play the classic remix of Steve Ballmer's "developers! developers!" whenever there are >=2 people in the room. On April 1st, of course.

cptaj• 2 years ago

I am, in fact, suggesting you do this.

alwayslikethis• 2 years ago

I'm not suggesting you don't do this

I hope this is intentional

eduction• 2 years ago

I love this company. Yeah!

urbandw311er• 2 years ago

Another state of the art people counting system would be ultra wideband. It can literally count the number of heartbeats in a room. See https://m.youtube.com/watch?v=adiUegDxZEs

davidteather• 2 years ago

UPL UPL UPL UPL

nicosalm2• 2 years ago

UPL!

zdw• 2 years ago

Back in the 90's we used finger - IIRC it would tell you the last machine someone logged into, which could be in the lab

Someone wrote a script to finger everyone in the entire CS department and tell when the lab was busy, by counting people logged in.

This work fine, except for on intro courses where some labs had lots of non-CS majors in them.

bks• 2 years ago

I had finger running on login to `finger stacy` I was at SDSU on a very large SunOS system and she was at a private school and I assume that computer was a bit more limited.

`Finger Stacy` would run every minute and typically be running for 15 minutes max... that is until I moved into the dorms and my machine was online all the time.

A few weeks go by and I get an email from the SDSU admin requesting that I stop fingering stacy as it was bothering the other Sysadmin. I remarked with a grin that all I was trying to do was in fact try to `name of the command` and they promptly deleted the script from the account.. It still makes me smile as I write this.

hotspot_one• 2 years ago

fdisk or mount?

neilparikh• 2 years ago

I built a similar system for my school’s CS club. I considered using a door sensor, but the eventual solution I settled on was a light sensor, because it’s almost always true for us that the lights are on iff the door is open.

This way, we don’t need to mount anything on the door, we just have a microcontroller plugged into one of the machines.

Our previous solution was a webcam that pointed to the lights that did a similar thing (implemented by someone before my time) but then it stopped working due to some driver issues, and I didn’t want to spend time investigating them.

_visgean• 2 years ago

In edinburgh we had this: https://devpost.com/software/the-marauder-s-app which showed who was using which computer based on logins.

withinrafael• 2 years ago

I saw "over-engineering" in the title and half expected to see Wi-Fi signals being abused to detect human bodies in the room.

https://arxiv.org/abs/2301.00250

TrackerFF• 2 years ago

My first thought was to put some motion sensors inside the room, along with a sensor for the door (open/closed), as well as some sensor for counting the people walking through the door. Maybe even some co2 sensor to detect changes.

zanchey• 2 years ago

The University Computer Club at UWA also has a door sensor!

https://door-status.ucc.asn.au/ucc

Runs off a Raspberry Pi with a bodged-up version of https://github.com/ide/pico-door-sensor/tree/main

This is at least the third iteration; some previous iterations are documented here:

https://www.ucc.asn.au/services/door.ucc

dbetteridge• 2 years ago

Still works better than the uwa door card readers

mdberkey• 2 years ago

Love to see the UPL on hacker news :)

fartfeatures2• 2 years ago

Point the camera at the door, use the camera to detect if the door is open.

grahamj3• 2 years ago

Mount the camera on the door facing down and watch for a moving floor

fartfeatures2• 2 years ago

That's a clever solution that would use less processing power. A camera pointed at the door provides the benefit of being useful for other purposes e.g. person detection / recognition and standard security camera purposes.

changexd• 2 years ago

Very interesting and fun read! thanks for the post, when I saw the public restful api endpoint I immediately tried to see if I can make request just to see the "NICE TRY!" message, haha, then I kept on reading how you managed this issue.

kleiba• 2 years ago

This would be a nightmare to implement from a legal standpoint in Europe.

lormayna3• 2 years ago

Even if you collected aggregated data and not storing any PII or images?

preston4tw• 2 years ago

room presence detection has been a long standing challenge in the smart home / home assistant community. it's cool to see home assistant used and adapted for this use case.

zimpenfish• 2 years ago

Back in the man.ac.uk of the early 90s, there were no cameras or YOLO models but we still wanted to know when machines (especially the colour ones! LUXURY!) were available.

We just had "`rlogin` to every machine in the lab, run `who`, and collate the output". IIRC there was an early version written by 'flup that I extended with a tidier output (including an X11 window), auto-refreshing, and easier machine selection (eg. you could select rooms by name with regex filtering.)

Good times.

amingilani• 2 years ago

If you’re okay using Cloudflare tunnels, just use Tailscale to create a VPN instead.

q3k• 2 years ago

Another solution in the age of portable electronic devices: detect presence based on active DHCP leases. You can let then people register themselves by MAC on some web portal so they can be visible by name if they're comfortable with that.

lormayna3• 2 years ago

Or Bluetooth. There are some problems (like MAC randomization) to get a real tracking with BLE, but to have a rough estimation, it's a good technique. And you can do it with just an ESP32.

JohnBooty• 2 years ago

Hmmm... interesting. How would that work? Just count the number of BT signals in a room?

How many BT devices is the average student or faculty member toting around with them at a given point?

Could conceivably be 4+ with some combination of: phone, smartwatch, tablet, BT speaker, game device, etc.

QuantumSeed• 2 years ago

When I was in the CS department at NMSU, the Computer Science building was open to the public 24-hours a day. I suppose that's not commonly the case nowadays.

ericpauley• 2 years ago

FWIW the building from the article is open to visitors during extended working hours.

Source: Currently in the building.

IvyMike• 2 years ago

Back in my day we just looked at the ctime of /dev/mouse

mashygpig• 2 years ago

Glad to see the UPL alive and well!

aabbcc1241• 2 years ago

If the camera doing people count is not working accurately, how about point the camera at the door, to check the door status ?

Overengineering a way to know if people are in my university's CS lab

Related Stories