Computer Vision Pose Estimation Development

DeepX: Pose Estimation Technology Review

At DeepX we are constantly coming up with new solutions, which can meet various business needs. One of such solutions is the implementation of real-time humans pose estimation into products that may be further used in a variety of business and day-to-day activities.

What is a human pose estimation?

Human pose estimation is the process of accessing the position of human body parts in space based on an image or a video with further tracking. Usually, the body or pose estimation is done by identifying, locating, and tracking the number of key points on a given person.

There are 2 basic types of estimation: 2D (x,y coordinates)  and 3D (x,y,z coordinates) pose estimation. For real-time humans pose estimation 3D mode is more effective since it allows not only to detect and estimate the location of the key points of a human’s body in 2D space but also to detect and predict spatial positioning of a person.

Where can it be used?

Here are some of the most exciting examples of pose estimation technology in practice today:


Recently, due to the spread of COVID-19 lots of fitness centers bore thumping damages because of their temporary closure. Consequently, lots of people had to find new ways to maintain a healthy lifestyle while being locked down in their homes.

In order to meet the demand, various web applications have started to propose libraries with recorded exercises. However, the users are now facing another issue: “Am I doing it right?”. Our team is ready to provide the answer using real-time pose estimation.

How does pose estimation work in fitness?

DeepX: Fitness Pose Estimation Demonstration

When a person starts exercising in front of a webcam, the image is being ingested by a CV/ML model (human pose estimator in this case), and the coordinates of the main key points on the body are detected (wrists, elbows, shoulders, etc). This allows us to check how the distances and angles change during exercise.

These values are further compared with the ones from the previously processed physiotherapist video, thus allowing us to not only get the number of repetitions but also their position and timing accuracy and display them in the tracker in the web interface.

Pose Estimation Pros in Fitness:

  • can be used with web cameras and mobile devices
  • reduces risks of injuries during exercises
  • analyzes the accuracy of done exercises    

Hospital Care

Patients requiring 24/7 care and monitoring is an always-pending issue for healthcare facilities. Such monitoring is heavily affected by a human factor. The risk of unintentional mistakes is even higher when patients are asleep and a nurse doesn’t have the possibility to check them on time. 

The introduction of a computer vision model with human pose estimation is one of the most effective solutions. It can help to detect whether the person`s conditions are satisfactory or whether there are some issues requiring the attention of hospital staff. 

How does pose estimation work in Hospital Care?

Computer Vision Healthcare Pose Estimation Example

The automated monitoring system includes a corresponding computer vision model (performing patient’s state detection and pose estimation) and a camera, which is placed in the medical ward. When the system detects that the pose of a sleeping person (or general patient’s behavior) looks risky for one’s health, it sends a signal to the control panel, where the information on possible risks is shown. 

Pose Estimation Pros in Hospital Care:

  • one nurse can monitor more than one patient simultaneously more efficiently;
  • reduces risks of medical care and emergency help provision untimely.

Industrial Safety

Real-time human pose estimation can become an essential part of the industrial safety system. As a rule, the problem of industrial injuries is directly connected with safety rules being ignored by employees.

Consequently, the introduction of instruments, which can track when employees break safety rules not only can reduce the number of industrial injuries, but also show respective gaps in the industrial safety system.

How does pose estimation work in Industrial Safety?

Placed around the industrial area, cameras with computer vision models are detecting employees and their body positions. Such a system can track body movement patterns and estimate whether such moves comply with the industrial safety rules or not.

If necessary, the computer vision model can also be configured to detect certain objects (such as tools, protective outfits, helmets, etc.), and check if the employees adhere to security measures. If an employee does not wear a protective outfit or performs actions that break security rules, the system will detect them and produce a corresponding alert.

Being situationally aware, the system can not only log the events for further punishing employees but also prevent dangerous situations by predicting them and giving a respective alert.

Pose Estimation Pros in Industrial Safety:

  • reduction of industrial injuries;
  • detection of gaps/flaws in industrial safety system;
  • reduction of company expenses in case of industrial injury being a result of security system flaws.


To sum up, real-time human pose estimation can be used in different spheres of business, where monitoring of human activity is needed. Apart from described spheres, this technology can be also used in such spheres as security, transportation, service industry, etc. 

Our team is ready to propose a unique, innovative, and horizontally scalable system based on specific company needs, regardless of the sphere of business.

If you are looking forward to implementing an AI-based video intelligence solution for your business, get a free quote from computer vision experts.

At DeepX, we synergize our multi-year expertise to dive deep into each business’ specifics and achieve measurable business benefits. With our continuous research and a history of successful computer vision software development projects, we offer highly specialized expertise in some of the most challenging fields of computer vision.

Close Bitnami banner