MACHINE VISION NLP TEAM JOBS CONTACT IMPRINT &
PRIVACY
DEUTSCH
Machine Vision
Due to advances in the field of machine learning in recent years, any pattern in image data visible to the human eye can also be made visible to a machine. Sometimes machines can even be trained to detect structures not visible to humans.
Criteria to discover attractive process automation projects, where visual information play a crucial role:
In our experience, only by combining knowhow of internal operations with machine vision expertise, projects can be framed well. Feel free to approach us with questions, especially whether we deem your project to be technically feasible.
Contact us
Use Cases in Machine Vision
Error detection

#machine vision #production #logistics

Defective components can cause major production losses. To prevent this, an algorithm can recognize patterns and distinguish faulty parts from faultless ones at an early stage.

#machine vision #production #logistics

Defective components can cause major production losses. To prevent this, an algorithm can recognize patterns and distinguish faulty parts from faultless ones at an early stage.

Defective parts are the major reasons for expensive factory downtimes.

A camera in the production process could take standardized pictures of car parts and Machine Learning algorithms could identify defective parts at an early stage.

Defective parts are identified before the car part arrives at the production stage and prevents expensive downtimes.

Travel-Ad placement

#machine vision #ecommerce #travel

A photo of an object is taken to simplify the process of creating the advertisement. This picture gets analysed and the finished advertisement is displayed to the customer.

#machine vision #ecommerce #travel

A photo of an object is taken to simplify the process of creating the advertisement. This picture gets analysed and the finished advertisement is displayed to the customer.

If users want to offer their house, car or other items for sale on online platforms, the process of creating an advertisement should be as simple as possible. In this example, the user takes a photo of his house.

The algorithm analyzes the images and automatically enters data that can be recognized on the image. If necessary, text files can also be used, e.g. a previous advertisement of the house.

The advertisement is displayed to the customer and the customer can edit the advertisement manually. This saves a lot of time and details of the advertisement can be checked automatically.

Analysis of handwritten documents

#machine vision #nlp #financial industry

Handwritten documents can be read out and prefilled using machine learning algorithms. By further input, these documents can be edited again, which again benefits the accuracy of the algorithm.

#machine vision #nlp #financial industry

Handwritten documents can be read out and prefilled using machine learning algorithms. By further input, these documents can be edited again, which again benefits the accuracy of the algorithm.

In many industries, documents are filled out in handwriting by own employees or partner companies. These handwritten documents often have to be typed in order to process them further.

With Machine Learning algorithms, patterns in handwritten documents can be analyzed. The more documents are available and the more uniformly the documents are filled out, the better the algorithm recognizes the contents of the handwritten document.

The pre-filled documents are transferred to an interface and can be edited again by employees. The algorithm learns new patterns and can apply them to new cases.

Analysis of X-ray/MRI images

#machine vision #health

Diagnostic suggestions can be made by analyzing X-ray or MRI images using an algorithm.

#machine vision #health

Diagnostic suggestions can be made by analyzing X-ray or MRI images using an algorithm.

In radiology, MRI and X-ray images are taken for diagnosis.

Machine Learning algorithms recognize patterns in the images and can (with sufficient amount of X-ray/MRI images) and make diagnostic suggestions.

Doctors automatically receive suggestions for diagnoses and can use their personal impression to clarify and verify the diagnosis.

Determining the price of a building insurance

#machine vision #financial industry

The optimum price of a building insurance can be determined by the customer specifying the address of a building, for example, and any existing pictures of the customer.

#machine vision #financial industry

The optimum price of a building insurance can be determined by the customer specifying the address of a building, for example, and any existing pictures of the customer.

The price of a building insurance depends on the location, type of building, number of floors and other factors. When analyzing the price, the customer's address is requested.

With the customer's address, the location and house are analyzed with satellite images and pictures of the customer and checked for pre-defined factors (number of floors, type of building, etc.).

The algorithm checks the properties of the house and generates the optimal price for this building. This price can be used, for example, to optimize the offers in price comparison pages.

Classification into quality classes

#machine vision #production #logistics

Through standardized images of different leathers, a machine learning algorithm can classify leather and assign it to the right garments.

#machine vision #production #logistics

Through standardized images of different leathers, a machine learning algorithm can classify leather and assign it to the right garments.

In the production process of clothes, materials are delivered by suppliers. The materials might have different qualities and these different qualities might be used for different clothes.

A camera can take standardized images of the leather surface and Machine Learning algorithms can learn how to classify leather into different quality classes.

Depending on the defined quality criteria the leather quality is determined and is associated to the related cloth production process.

Trendscouting

#machine vision #ecommerce

An algorithm can analyze thousands of influencer styles and outfits and predict next season's trends.

#machine vision #ecommerce

An algorithm can analyze thousands of influencer styles and outfits and predict next season's trends.

Thousands of styles, outfits and items of influencers are analyzed for trend scouting in order to identify the trends of the new season early and to be able to reliably forecast the demand for "must-haves”.

Machine learning algorithms analyze images from social media channels (Instagram, Snapchat, Facebook), fashion magazines and other sources to check for defined "properties" such as color, patterns and material.

The algorithm develops the "trend probabilty" for individual dressets based on "properties" and trend score of influencers on the basis of their categorization.

Fraud detection in claims processing

#machine vision #nlp #financial industry

Neural networks are used to detect and filter out patterns of fraud cases. Conspicuous damage reports are reported to claim handlers and checked manually.

#machine vision #financial industry

Neural networks are used to detect and filter out patterns of fraud cases. Conspicuous damage reports are reported to claim handlers and checked manually.

Customers send insurance claims reports as text documents or pictures. Fraudsters assume that particularly low-value damages are not thoroughly examined.

Neural networks are used to identify and filter out patterns of past fraud cases or accumulations of conspicuous, current damage reports.

The selected damage reports are reported to the claim handlers and checked manually by insurance fraud specialists.

Location analysis

#machine vision #real estate

A machine learning algorithm analyses the properties and historical and current data of a property. This results in a score by which the attractiveness of an object can be evaluated.

#machine vision #real estate

A machine learning algorithm analyses the properties and historical and current data of a property. This results in a score by which the attractiveness of an object can be evaluated.

In the real estate sector, assessing the attractiveness of the situation remains a key success factor.

Machine Learning algorithms can help to assess attractiveness by evaluating historical sales and offer data and automatically analyze the proximity to highways, public transport, popular restaurants, residential areas and shopping opportunities using satellite images.

Taking into account the pre-defined criteria and historical data, attractive properties and locations for rental, sale or project development are identified and a "score" for assessing their attractiveness is developed..

Classification of equipment damage

#machine vision #production #logistics

Photos of defective spare parts are analyzed by a machine learning algorithm and the damage is classified. This information is used in the development of new products and in the selection of suppliers.

#machine vision #production #logistics

Photos of defective spare parts are analyzed by a machine learning algorithm and the damage is classified. This information is used in the development of new products and in the selection of suppliers.

In the development of new products, it is important to identify and eliminate the defects of past product lines. To ensure this, the service technician takes a photo of defective spare parts.

Based on historical data and similar photos of this spare part, the Machine Learning algorithm analyses the damage and classifies it into pre-defined classes.

Based on the classification of repair damages from previous product series, detailed information on repairs and complaints is available. This information is incorporated into product development of new products and the selection of suppliers and components.