How can you deploy the PPE detection solution?

It's easier than it may seem. We have prepared the answers for the frequently asked questions about this solution.

You seem to have a solution for my company. How do I get started with it?

Welcome, dear guest! Our team aims to offer responsive, competent and excellent service. So first of all let us invite you to an intro-call, where we can dive deeper into your business objectives and figure out the way we can provide you the best solution. Feel free to ask any questions, we’ll do our best to help you.To request a free, however full-fledged consultation, proceed here.

What steps will be taken from now till the full deployment? How long will it take?

Step 1: Fill in the contact form.

We’ll greatly appreciate it if you describe your challenge in a few words so we can prepare an individual presentation for you.

Step 2: Intro-call with our business development representative (approximately 1 hour)

We’ll provide you with a laconic presentation, sharpened to your request. Then we’ll get through your business objectives and choose the best option for how to get forward. There you can ask any questions concerning our solutions, approaches and services. No worries, you don’t pay for consultation!

We may take several iterations of discussions, product presentations and consultations with your team and our specialists. All this is intended to offer you the best solution for your individual business objective. We want to make sure that your request is completely satisfied and oriented on the best performance.

Step 3: Discovery workshop (1 week)

On that stage we define the list of protective clothing and equipment for monitoring and which cameras will be used for the pilot. We’ll need to have an appointment of the employee, responsible for safety control, to whom notifications about violations will be sent.And, surely, together we will establish the success criteria for the pilot project.

depositphotos_80434210-stock-photo-websi

Step 4: PoC Project (3 weeks)

Step 5: Deploy in production and scaling if needed (from 1 week)

After the workshop, we run a pilot project before scaling on the full worksite and all the areas. Here we analyze the outputs, and, if needed, collect additional data to make the model fit your special conditions. Then we proceed to creating an action plan for future implementation and scaling.

Step 6: Training and consultations (optional)

We can train your staff to work with the system. Moreover, we will provide consultations at first after launching the entire system in production.

Step 7: Extended support and customization (on demand)

Together we may plan the scope of support , as it should comply with your individual needs and objective. Solution customization is also an option if you need some special functions or capabilities.

What deployment options are available?

PPE Monitoring Platform is available by subscription in the following options:

Option 1: SaaS

SaaS product at https://ppe.vitechlab.com/. This option provides the fastest start, as you just need to connect your cameras and start using the solution.

Option 2: Cloud

This is also the way to start quickly, and fits you if you need to take into consideration the specific security requirements, and use your private cloud. The solution is ready to use and is deployed in the public cloud to minimize IT infrastructure costs and to accelerate time to market. Subscription can be issued through AWS Marketplace or direct contract minimum for 1 year.

Option 3: On-premise

This variant fits you, if you want a solution to work locally, and you have the appropriate server hardware for the platform launch and performance. On-prem deployment is for organizations that have to comply with special security rules, or for those that need to operate offline most of the time. For on-premise usage, the PPE Monitoring Platform is provided as ready-to-use Docker containers with special licenses monitoring scripts and requires minimal access to the Internet. Minimum contract - 1 year.

Option 4: Hybrid

Hybrid deployment (on-prem IT infrastructure plus public cloud) suits organizations that work with low-bandwidth internet connectivity OR have spikes in processing that cloud will smooth out.

I might need customization or support. What about this?

We understand that turbulent markets may establish new priorities for business as the time goes by. Moreover special functions of a product can expand business opportunities and hence provide a significant competitive advantage. Therefore our team is ready to serve your company and provide consistent solutions that fit your goals.

Our Advanced Support Services include:

system monitoring
collection of the additional dataset
model retraining for enhancements once every 3 months
hot and priority fixing bugs and system problems
advanced training.

If any additional options are needed, it's all up for discussion. Contact us anytime and for any questions.

Some examples of possible customization:

Adding your custom PPE items to be monitored

Integration of the solution with with you systems (access control, social distancing, etc)
Adding some new functions and capabilities (color detection, size detection, etc)

Or you can request an individual customization, and we will do our best to meet your needs.

How does the solution work on AWS Marketplace?

The PPE Monitoring Platform consists of two components:

The ML model itself, which analyzes images from live stream from client’s surveillance cameras on the worksites;

The Web interface, which transfers images from cameras to the ML model for analysis. The interface is user-friendly, so it serves as the main workplace environment for an end user. It provides an ability to:

control the cameras;
observe the process of video stream analysis in real-time;
get and configure notifications about detected non-compliance cases:
get access to all the detected violations, automatically saved to an archive;

manage users and accesses.

The architecture on AWS Marketplace is depicted below:

Have any questions? We are here to help.

What are the technical requirements for an ML model server and web-interface server?

For ML server:

Server with Docker and NVIDIA drivers installed
GPU: NVIDIA K80/P100/V100/A100/T4 or similar
CPU: Intel Xeon min 2.0 GHz
RAM: min 16 GB
Storage: min 100GB
EDGE platforms like NVIDIA EGX A100 or NVIDIA Jetson

For web-interface server:

Storage: min 1TB (SSD)
CPU: Intel Xeon min 1.3 GHz
RAM: min 8 GB

What are requirements for cameras and monitored objects?

Please note, that VITech Lab does not provide the needed hardware, we use your existing cameras. If this is a kind of a barrier, we have partners who can provide you with cameras.
Resolution - from 2 MP
Video resolution - from 1920 × 1080 pixels
Video Format - H.264
Max FOV: 80 degrees
Color
The size of the analyzed object - a minimum 5% of the camera image size

What are requirements for cameras and monitored objects?

The ML model speed is up to 20 fps (V100-based server). It also depends on the cameras’ image resolution and the number of people on each image.

Another important thing about speed is that 1 server can work with 10 cameras in real-time, while one stream processing runs at a speed of 10 fps (V100-based server). The number of processed threads can be increased by reducing the processing speed.

How do you guarantee security for my data?

If you choose to deploy the solutions in cloud, using your AWS Marketplace account, then be sure that AWS Marketplace itself meets high security level and has the appropriate certificates. In such case we do not have access to your data. For more details visit: https://aws.amazon.com/en/compliance/, https://aws.amazon.com/en/security/

If you choose an option where the solution is deployed locally, then all your data is kept locally, and we also do not have the access to it.

Moreover, we sign an NDA agreement, so as to secure all possible informational exchanges, data and terms of cooperation.

Need more information?

Get free consultation