Cameras as a sensor: real-time analysis of video images

Skip to Content

Image and video analysis

How AI-aided machines learn to see

Article: Cameras as a sensor: real-time analysis of video images

Since 2018 we have been working on AI-based analytics for images and videos in the vsion.ai corporate venture. With our expertise in machine perception, we have been working with our partners ever since to meet your challenges in rail operations.

Systems that monitor environments or control automation are only as intelligent as the available data allow. Using conventional sensor technology, digital systems are only able to capture the environment or even understand the context to a limited extent. A trained eye, on the other hand, recognises the situation at first glance. This is where vsion.ai's self-learning analysis based on image and video data steps in. Supplementary sensors transform the system into a powerful overall package.

"We solve real problems for our partners using machine perception. That means we not only tackle complex technical challenges with our expertise in machine learning and artificial intelligence but also develop a deep understanding of the specific context in order to offer our partners in the Group a stable and productive system that not only makes their actual work better or easier but is also fun to use."

Avishek Sen-Gupta, AI engineer, DB Systel

Profile picture Avishek Sen-Gupta

Making better decisions with new insights

Self-learning image analysis enables a wide range of applications in the railway environment. Combine existing or new camera systems with intelligent analysis and optional additional sensors. Automatically detect and analyse conditions, objects or specific patterns and processes. We can work together to integrate these analyses into your existing applications or make them available to you via your own user interface tailored to your needs. If required, the image and video data are anonymised in compliance with GDPR.

How Deutsche Bahn benefits from intelligent image analysis:

Learning to distinguish the important from the unimportant

Nowadays, cameras are used almost everywhere, in regional trains, suburban trains, at stations, in maintenance depots or through the use of drones. The trick here is to extract truly important information from the myriad of images. This is exactly where vsion.ai comes in with IDA for the digital evaluation of railway-specific issues such as train exteriors, train interiors and track infrastructure.

Gaining new insights

Intelligent image analysis offers many possibilities in DB's rail operations for optimising processes, detecting defects or making more informed decisions. It can automatically detect vandalism, dirt and objects, count passengers or optimise cleaning and maintenance cycles.

Rapid potential and feasibility analyses

We combine our expertise in machine perception with our knowledge of the railway. In order to be able to identify potential quickly, we provide advice on possible applications of AI-aided image analysis in the respective area of responsibility. This is done in short iterations in the context of proof of concepts. We provide support from consulting, hardware procurement, data processing and data storage through to data provision, depending on the use case.

Using visual data productively

Automatic analyses, real-time notifications, clearly structured dashboards and APIs map all events in an easy-to-use form. IDA maps the entire processing chain of visual data: from perception and data processing through to analysis and preparation for presentation.

IDA: Transform your environment into structured data

IDA is intelligent digital assistance for the analysis of image and video data. Automatic detection of objects or conditions helps to gain new insights in everyday work. In practice, IDA is already being used to monitor train exteriors, train interiors or track infrastructure or to provide security at railway stations. Data collected via camera sensors can be made accessible via our own frontend or directly integrated into existing systems. For example, IDA notices soiling such as graffiti, counts passengers or detects missing or non-compliant components and reports corresponding events according to your requirements.

If the aim is to use automated sensor analyses in your own business area,  we will be happy to advise you.