Steadforce at the TDWI 2021

Talk about distance measurement in Computer Vision

Computer vision, one of the hottest areas of artificial intelligence, is taking over the 'visual world'. At TDWI, we discussed in a talk how it is possible to teach a camera the capabilities of an eye. How does the computer interpret and understand the digital images to identify and classify objects. Meanwhile, computer vision is taking on more sophisticated tasks. Thus, competing alongside human skills, for similar or even better results in these tasks.  For example, diabetic Retinopathy can be diagnosed using computer vision (equivalent to a trained physician) with a sufficiently large amount of data.

In this presentation, we also discussed in detail the current developments and future prospects of Computer Vision.

Referring to the different use cases and their impact on the real world in different problem areas , such as:
‍

1. Object Detection
   Provides the ability to simultaneously recognise hundreds of categories of different objects in an image. Enhanced with the ability to segment individual image areas with pixel precision.
   
   
2. Gesture / Action Recognition
   Means the recognition of patterns in an image sequence in order to identify actions performed with it.
   
   
3. People Tracking
   A very interesting example of this is AmazonGo stores in the US which not only keep a track of what the users buy but also use what they don’t buy in order to analyse consumer behaviour.
   
   
4. Facial Recognition
   Mainly used in surveillance cameras and security systems. It can be used to identify people to support police work. It can also be used to check identity in payment portals or to unlock one's own phone.
   ‍
   
5. Autonomous Driving
   Computer vision is a critical technology for self-driving vehicles. It must be able to detect and assess the environment very accurately so that the vehicle can react correctly and safely to different traffic situations.
   
   
6. Eye tracking
   Is the tracking of view position to determine the corresponding user behaviour. Used, for example, by film directors to analyse the actual focus of the audience on their film.
   
   
7. Medical Imaging
   The importance of computer vision is growing significantly in the medical industry. There are already applications for diagnosing skin diseases with the help of a camera. The resulting medical data can be analysed precisely with the help of computer vision.
   ‍
   
8. Precision Agriculture
   To make farming more efficient, the fields are subject to constant monitoring. This makes it possible to look for signs of diseases or pest infestation of the plants. In this way, if problems arise, they can be eliminated quickly and without major consequential damage.
   ‍

Further, we looked into a very small application area, i.e. distance measurement. To measure any distance from the camera, we need the depth.  Which can be captured using Lidar or Time of flight sensors, Stereo Depth sensors, Structured or coded light cameras, or just another RGB camera with some pre-determined marker information.

This can be very important in different industries like Virtual- or Augmented Reality where we try to reconstruct a real world in which, for example, virtual objects have to be placed at certain positions.

Or another example could be Robotics. The issue here is to understand the surroundings for Navigation, mapping, collision avoidance and detecting the size of approaching object. After discussing the Use Cases, we also see a simple demo using Python.

We demonstrated  how we can use distance measurement to comply with the Corona measures in the workplace. By creating an inventory of the chairs used and then disinfecting them.

These examples also illustrated various practical applications of Computer Vision in today's world.

The "Object Detection" application area is covered in detail in our blog post "Explainable Object Detection - a practical How To".