4 Computer Vision Apps for Health and Safety on Your Construction Site

By Liz Oz • Nov 05, 2021

Computer Vision for Construction Safety Solutions

Injury rates in construction sites are 71% higher than injury rates across all industries on average. Construction can require hundreds of people laboring with heavy equipment or at unsafe heights. Construction, by definition, has inherent health and safety risks, so injuries can be challenging to reduce entirely. 

However, construction managers know the value of a safe workplace. When injuries go down, productivity improves and companies see changes like increased employee satisfaction and workers’ compensation expense reductions. 

‌Traditionally, organizations rely on company culture and employees to do health and safety monitoring. However, this isn’t always effective due to human error and lack of knowledge. Even if you host frequent training sessions or safety meetings for your employees, there's no guarantee they will remember or understand everything you said. Even if they remember and understand, there’s no guarantee they will implement these policies.

That's why some tech-savvy construction companies have started adopting computer vision (CV) to ensure employees are operating safely. CV digitizes the physical environment to help identify situations, behaviors, and circumstances that may be unsafe. For example, CV can identify a situation like a truck containing chemicals coming into a work zone as potentially dangerous. From there, employees can schedule necessary safety support measures, analyze what may happen, and try to reduce risks.

Computer vision can be construed in certain organizations as a form of high-tech surveillance. While the technology does have the power to track people and movement, computer vision does not store or share information about an individual’s movement patterns or actions taken with anyone. It’s important to understand that CV gives overseers more visibility into the overarching effectiveness of their safety programs, while protecting employee privacy. Computer vision is not intended to be used for strict employee monitoring. 

In other words, computer vision helps people understand things that happen in a physical environment. It allows for predictive analysis, which can help you know what may happen in the future based on past activity patterns. Furthermore, it enables you to monitor what you usually can’t put sensors on in your physical environment. For instance, you can’t put sensors on employee's hats to ensure they’re working within a safe proximity of one another. However, you can install a set of CV-enabled cameras that calculate the distance between each worker. These cameras could also help you to determine:

  • If someone has overextended a ladder
  • If a truck is going too fast
  • Other safety issues that you may not be able to determine if you use traditional methods like sensors or in-person monitoring

Read on to learn about four real-world CV applications for construction safety.

1. Object Detection Alerts for Safety Gear and PPE

CV can send you alerts when a particular object is, or more importantly is not, present. This CV application can be beneficial in detecting employees not equipped with proper PPE like helmets, vests, face shields, and spectacles. 

You can also train CV models to detect any of the following:

  • Whether an employee is wearing earplugs or earmuffs in a noisy area
  • If gloves are being worn when operating different types of machinery
  • If masks and respirators are being used to protect against exposure to chemical fumes or pollutants

In theory, any safety equipment you want to enforce can be monitored with computer vision. CV models can be trained to detect the presence of desired PPE, and can even become trained well enough to detect the minute differences between types of masks, eyewear, or gloves needed for different types of jobs.  

Now, it’s essential to understand that outfitting one camera on a job site to monitor hundreds of employees’ PPE compliance is likely insufficient. Edge computer vision, which connects to your pre-existing security camera infrastructure, relies on strategically placed equipment pointed directly at key interaction points to work appropriately. For example, if you have a hazardous area in your job site, having a dedicated camera for it is probably worthwhile due to the risk of injury. Similarly, if you have a sprawling worksite with hundreds of workers moving around, it’s best not to rely on one camera to detect PPE. One camera can only pick up details so well. Training more cameras on smaller areas leads to maximum accuracy.

2. Spacial Awareness with Person Detection Models

Workplaces need computer vision to constantly monitor physical distancing and send alerts to notify leadership of employees not in compliance. The last thing you need is a worker struck by a swinging backhoe through a lack of attention to detail. 

The application to do this is essentially a basic object detector with bounding boxes around each person in a video feed. While it can help solve simple problems such as determining how many people are in a warehouse at a specific time, it’s hard to determine how far apart each person is. That’s because this basic model only identifies the center of each person, not their height or size.

TIP: Many engineers make assumptions about height and room dimensions to calculate how far each person is. However, you need to add depth and perspective if you want precision by using RBG depth cameras. These cameras give you the pixels in an image as well as the depth of every pixel. You’ll also have the Z-axis, so you’ll be able to know the geometry of every object in view.

3. People Counting for Capacity Restrictions

To count how many people are in a given space for capacity restrictions, you can also use object detection models. 

If the space in question is vast, you can cut a broader view into specific zones. This will allow you to determine how many people are in one half of an area versus the other. A premier example of this is alwaysAI’s zone tool. With this tool, all you have to do is draw the zone on your camera feed to cut the view.

Since people are constantly moving around, you can also use people tracking. Also known as object tracking, people tracking goes frame by frame. As people move around a camera frame, the program pipes the object detection results into a tracker, assigning a unique ID to each person. Since each person in a room or area has a unique ID, the program can easily track the movement of individuals through different zones to understand the total capacity and which zone people are in. This level of tracking gives you enough information to act accordingly if any areas are reaching or have reached full capacity.

4. Human Activity Detection for Safety Compliance

Activity detection models can monitor for unsafe behaviors like: 

  • People running 
  • Workers slipping in an unexpectedly wet area
  • A ladder moving excessively as a worker climbs
  • People not holding handrails on stairs 
  • People entering dangerous or restricted areas

After you’ve gathered enough statistics about harmful "activities” at your site, take a look at the statistics compiled by your CV software. Suppose you notice particular patterns that need fixing (i.e 15% of the staff slipped in specific wet areas one day). In that case, you can dedicate resources to rectify those issues, whether it calls for additional training, corrective action, or future investment. 

To Wrap It Up

Without computer vision, you would have to rely on your company culture and employees to do the health and safety monitoring. 

Implementing a culture that encourages health and safety monitoring is essential but often difficult because it’s almost impossible to make sure everyone is on the same page. Employees have to understand and prioritize their company values and act on these values. This traditional method of ensuring safety policy compliance in your worksite is highly dependent on people paying attention, doing the right things, and knowing what the right things are.

That’s why you should use computer vision. CV will always protect your employees. Unlike humans, CV will never lose focus, fail to do the right things, or forget the task at hand. CV can also help you strengthen your company culture. So if you’re having problems with people not holding the stairwell in a specific area, for instance, you can use CV to determine if more training is needed.

CV will also help you comply with OSHA since it can keep detailed records of safety compliance. Keep in mind that CV is not about busting people for not following rules — instead, it’s about being able to train people better.


About alwaysAI®

alwaysAI® provides developers and enterprises a comprehensive platform for building, deploying, and managing computer vision applications on IoT devices. We make computer vision come alive on the edge - where work and life happen. The alwaysAI platform offers a catalog of pre-trained models, a low-code model training toolkit, and a powerful set of APIs to help developers at all levels build and customize CV apps. alwaysAI® has an easy deployment process and a state-of-the-art run-time engine to accelerate computer vision apps into production quickly, securely, and affordably.

Make advanced AI easy for your organization. Register for free to get started today, or schedule a demo and talk to our computer vision experts about your business. 

By Liz Oz • Nov 05, 2021

Developer stories to your inbox.

Subscribe to the Developer Digest, a monthly dose of all things code.

You may unsubscribe at any time using the unsubscribe link in the digest email. See our privacy policy for more information.

alwaysAI Ad
stylized image of a computer chip

Sign up today and start your project

We can't wait to see what you'll build!