2nd round of public voting is open. Vote for us on LinkedIn: https://www.linkedin.com/posts/vi-k-ly_check-out-the-final-four-projects-from-our-ugcPost-6676959483345846273-AuhT

Team: Vi Ly, Sagar Bansal, Moksh Nirvaan

Github URL: https://github.com/sagban/redisPPEScan

Does your app use Event-driven architectures or Redis Modules?

Event-driven architectures and Redis Modules

Describe Redis usage further:

A live camera stream is stored in the Redis Stream in the form of frames. After that, each frame is read by the RedisAI modules which sends the results back to the new Redis stream. This Redis stream is used to show the results to the control room.

As in the diagram above, Redisafe web-app uses Redis Streams to capture the input video stream: camera:0. RedisAI to classify the images using the TensorFlow mask detection model and we used Express.js to serve the client-side. Eventually, it forwards the classified images to a stream: results

Inspiration

What will be the working situation for medical staff in hospitals during and after the COVID-19 pandemic? How can medical staff quickly and securely log in and perform PPE safety checks while dealing with a huge influx of patients in critical conditions? How can our solution for hospitals later be scaled and implemented to be an essential tool for automating the daily operations at hospitals even after the COVID-19 pandemic is over?

To answer these core questions, we did some background research to identify the main challenges in order to develop the best solutions around those:

The problem with PPE safety check:

According to the CDC, during the COVID-19 pandemic, all healthcare workers must follow strict guidelines and protocols from OSHA regarding wearing PPE. PPE prevents contact with an infectious agent or bodily fluid that may contain an infectious agent, by creating a barrier between the worker and the infectious material. Gloves, protective hands, gowns and aprons protect the skin and/or clothing, masks and respirators protect the mouth and nose, goggles protect the eyes, and face shields protect the entire face. N95 masks are the PPE most often used to control exposure to infections transmitted via the airborne route. Therefore, checking the medical staff’s PPE safety protocol is especially crucial during this pandemic.

An innovative Redis-based solution:

To find the solution to this problem, we explored the open-source tools from Redis. After some exploration, we found that RedisAI and Redis Streams are the most suitable for our idea of PPE detection through live video footage from the webcam. This is how we came up with the web-application RediSafe. Using the power of Redis module for serving tensors and executing deep learning models, the web-app will primarily be used to check for medical staff’s PPE to see if the staff is following established safety protocols to minimize any exposures to the disease.

How we built it:

We have developed this application by using the Redis Streams’ event-driven architecture and Redis AI module.

Redis Streams

Alt text

Fig 3: Use of the Redis Streams

We have used Redis Streams for this project in the following ways: 1) camera:0 - To store the frames from the live camera feed. 2) results- To store the classified results from the Tensorflow model running on Redis AI.

Redis AI

Redis AI module is used to deploy and serve graphs (Tensorflow mask detection model) by leveraging Redis' production-proven infrastructure, as well as maximizes computation throughput by adhering to the principle of data locality.

Working

Alt text

Fig 4: Deployment process using Redis

A live camera stream is stored in the Redis Stream in the form of frames. After that, each frame is read by the RedisAI modules and send the results back to the new Redis stream. This Redis stream is used to show the results to the control room.

As in the diagram above, Redisafe web-app uses Redis Streams to capture the input video stream: camera:0. RedisAI to classify the images using the TensorFlow mask detection model and we used Express.js to serve the client-side. Eventually, it forwards the classified images to a stream: results

Technical Requirements:

The packages required for this project are as follows:

Redis stream

OpenCV

Redis - python client

RedisAI - python client

Numpy

Tensorflow

Accomplishments that we're proud of

We managed to finish the project in such a limited time of 2 weeks in our free time from school and work to learn new tools from Redis system. We still keep striving to submit on time while learning and developing at the same time. We are really satisfied and proud of our final product for the hackathon.

What we learned

Through this project, we learned to implement a complicated image-recognition deep learning modelsRedis. We also learn the process of developing a mini data science project from finding dataset to training the deep learning model and finally deploy & integrate it into a web-app. This project can’t be done without the efforts and collaboration from a team with such diverse backgrounds in technical skills.

Alt text

Fig 4: Developing team

What's next for RediSafe:

In the next 2 months, our plan is:

  1. Through our PPE checking, we will collect a large amount of data from the users. In our next step, we will explore and utilize Redis Labs to store the collected data in the database management and memory caching process. We will also use the Redis Timeseries to aid our data analysis process from the information collected from live video.

  2. We also want to host our web-app on the Redis Cloud Essentials with the free credits provided.

  3. In the long run, we will work on raising funds to invest more in the R&D process.

  4. We will partner with research labs to collect more datasets and find hospitals to test our solutions.

  5. Regarding our R&D, we plan on improving the performance of the platform, preferably by reading more scientific literature on state-of-art deep learning models implemented for PPE recognition.

  6. We want to develop this into a full-scale tool with many other functions so it can become an integral part of security check-in at hospitals even after pandemics. One of the things we want to add is the analytical tools which use the Redis TimeSeries module to track a count of the number of people screened.

Built With

+ 106 more
Share this project:

Updates