Fall 2020 Design I Final Competition

Section T: Replace with Team Name

Problem Statement: How might we help small businesses conform to COVID-19 induced capacity guidelines?

Team Members: Dorian G.Cauwe, Conor P. Haithcoat, Vincent Denley-Spinola, Henry D Upton, and Jackson T. Wells

Instructor: Donna Bodeau

12 Comments

  1. Great and creative use of video to present your ideas. A few questions:

    1) Do you feel that a small business person without much technical knowledge can install and service this product?
    2) Could there be a market beyond small businesses? For example, could this counter help Mines let students know how many people are in Periodic table or in the library by feeding into the M app, for example, so they may decide to go elsewhere? Are there other uses beyond the example you used in the video that could keep this product going after the pandemic is over?
    3) How sensitive is it to group? You showed at most 2 people at a time, but if larger groups enter and exit can it detect those numbers quickly?

    Great job!

    -Peter

    • Hi Peter,

      Thank you for your questions!

      1) To address your first question, we tried to make it as easy as possible for small businesses to install and maintain the product. Because the system attaches magnetically to the top of a door frame, installation is as easy as clipping the counter to the doorway. The system does need a constant power source, so businesses do have to find a way to thread a wire to the system, probably through an electrician or other professional. However, this a fairly routine and inexpensive job, as many smart doorbells and security cameras also need a constant power source to operate. Very little maintenance should be necessary to keep the product operational. The product can easily be monitored through the online control panel in case something does go wrong, and a list of common problems (including a dirty lens, faulty power supply, etc.) will be included in the control panel so businesses can find out how to fix the issues easily.

      2) There is definitely a market beyond small businesses. Because the counting algorithm is already in place, it can also be used as a general-purpose counter in the situations you have described. The counters can currently all be connected to a local Wi-Fi network, so it would be fairly easy for the system to interface with any other applications. We have built a basic Application Programming Interface for other apps to talk to the counter, although this API is still in its early stages.

      3) Our pre-trained detection model runs in constant time, so the time it takes to process an entire frame does not depend on the number of people inside of it. We would theoretically be able to detect as many people as can fit inside the frame, but we haven’t had the chance to test with more than about 2-3 at the same time. Since the final algorithm runs at about 20-30 frames per second, we also wouldn’t miss any rapid motion.

      Thank you!
      – Replace with Team Name

  2. Awesome job on the video and prototype.

    What other project ideas did you explore before deciding to pursue the capacity counter?

    Are there any risks associated with your solution? Are there any privacy concerns identified in your stakeholder engagement?

    Can you give me a breakdown of the monthly subscription fee? What if customers want to house their own data?

    How is your solution unique from other options on the market? (lots of apps and products when i google Capacity Counter)

    Thank you,
    Garrett Erickson

    • Hi Garrett,

      Thank you for your questions! I can answer question three.

      The monthly subscription covers the costs of hosting data from the counters (video + counts), and is roughly based on current storage prices. We thought it useful if customers could view video footage of motion events from their capacity counters through the settings page, since that could be used for purposes beyond counting (like security). We estimated that a maximum of about 400-500 GB of data would be generated from recording motion of 2 counters (the average needed by our stakeholders), which would cost about $10/month on Amazon S3. The rest of the costs cover access to additional online dashboards that can provide historical data so businesses can better understand the traffic patterns of individuals entering and exiting the establishment, as well as the development costs needed to maintain the online dashboards and provide security updates and support for the counters.

      We didn’t mention this in the video, but there is no requirement for small businesses to store their data with us.

      If the small business only needs current counting data, then the amount of data stored would be very limited, and there would be no need for an online server. We have also started working on a minimal Application Programming Interface for small business to access their own live count data, since we identified that as a stakeholder need.

      Thank you!
      – Replace with Team Name

    • Hi Garrett,
      As far as your second question goes, there are not really any privacy concerns that come with our product. This is because we are not actually storing any data of the people that come in and out of the store. With that said, there are still a couple of risks that we have come up with. The first is that there is a chance that someone could hack the camera, although since there would be no data for them to try and steal, we deemed this risk fairly low. The second risk that we came up with was the chance that it might fall on someone. This product is positioned right above the door frame, so if it fell, it might hit someone. But we came to the conclusion that because the product is light, and has a case around it, if it did fall, there would be no significant damage to either the camera or person that it might hit.

      Thanks
      – Replace with Team Name

      • Yes, even though we do record data from the cameras and companies can store that with us, we don’t individually identify anyone (or store data on them). Similar to how a security system operates, no one other than the business has access to that data.

    • Hi Garret,
      I an answer your fourth question about existing market solutions. There are a few market solutions that are similar in concept and function to the capacity counter. An example of one of these systems is called Traf-sys. Traf-sys, like our camera count system, is designed to count people moving in and out of venues, and keep track of capacity for business owners. Just like the capacity counter, Traf-sys uses an algorithm to track incoming and outgoing entities, and uses an online interface to allow users interact with the data Traf-sys collects. The key difference between our system, Traf-sys, and other similar systems is the capacity counter’s accuracy and its ability to consistently distinguish between human and non-human entities that enter a venue due to the combination of our infrared system and our algorithm, where Traf-sys and other systems cannot.
      Thanks for your questions,
      Replace with Team Name

    • Hi Garrett,

      Thank you for your questions! I can answer question one.

      When we were brainstorming we actually started at trying to reduced PPE waste from masks and other COVID related protection gear. Through some stakeholder engagement and research we figured out it was not as big of a problem as we thought. Then we found we wanted to do something with tracking the capacity of businesses. We all research different ways of doing this such as a laser tripwire, a button based system, a regular camera, a pressure sensor, and an IR camera. We decided on the IR camera system due to the accuracy and pricing of it. The IR camera is one of the most accurate ways for us to track capacity for only a couple more dollars than a regular camera. The IR sensor is also the best at distinguishing multiple people walking through the door at once. I hope this fully answers your question.

      Thank you!
      – Replace with Team Name

  3. Great presentation, team! I enjoyed the video and the content of your presentation.

    Few questions:
    1. How is the device able to distinguish who is coming in and who is getting out of the respective location?
    2. Once capacity is reached, how will inform the operator?
    3. How does your product compare with existing market solutions?

    Thank you, Dr. H

    • Hi Dr. H,

      Thank you for the questions! I will answer question one.

      The path tracer algorithm traces a line between different frames from the Capacity Counter’s camera. If any line intersects with the counting line (the green one in the middle), then it knows that someone crossed the line, but not the direction. The direction is identified by looking back as far as possible to the first time that person entered the frame. If the person was first sighted at the top of the frame, we know they are coming in to the store, if the they were first sighted at the bottom of the frame, we know that the person is exiting. This can be inverted in the settings if the counter is placed inside of the store, in which case leaving and exiting would be inverted from the counter’s POV. Since it is also possible for person to move back and forth across the counting line without leaving the counter’s “view,” the algorithm only looks back at the previous time the customer crossed the line if they’ve crossed it before (not to when they were first sighted), and inverts their last move (example: if last time the customer exited, they must have entered now).

      Thank you!
      – Replace with Team Name

    • Hi Dr. H,
      I can answer question number 3 for you. There are a few market solutions that are similar in concept and function to the capacity counter. An example of one of these systems is called Traf-sys. Traf-sys, like our camera count system, is designed to count people moving in and out of venues, and keep track of capacity for business owners. Just like camera count, Traf-sys uses an algorithm to track incoming and outgoing entities, and uses an online interface to allow users interact with the data Traf-sys collects. The key difference between our system, Traf-sys, and other similar systems is camera count’s accuracy and its ability to consistently distinguish between human and non-human entities that enter a venue due to the combination of our infrared system and our algorithm, where Traf-sys and other systems cannot.
      Thanks for your questions,
      Replace with Team Name

    • Hello Dr. H,
      Thanks for your questions. For the second question, we have a couple of things for how it will inform both the operator and the general public. For the public that are looking at the LED display in the window, the numbers will change color based on how full the store it. For example, if a store’s capacity was set at 20, and only 3 people were inside, the numbers would be green, if say, 15 people were inside, the numbers would be yellow, and if the store was at or over capacity, the numbers would be red. We also have speakers inside the housing with the camera that would emit a sound when capacity is full. The sound combined with the red numbers will let people know to wait before coming inside.

      As for letting the operator know, we have an app that employees could use to see how many people are in the store at the current moment. They could also change the capacity limit through the app if they choose to.

      Thanks,
      – Replace with Team Name