Feature Recognition

Increase the efficiency of your field surveying by implementing the appropriate AI algorithms to detect a wealth of information from images in an automated way.

Automatic feature recognition for increased efficiency

Implementing the appropriate AI algorithms enables the automated extraction of information from aerial imagery. This information, which can be about building characteristics, land use, vegetation, or vehicles, not only provides you with insights into the exact location of your objects, but also gives you information on what these objects look like or what is happening in a certain area. This allows you to limit the execution of your field surveying and inspection tasks to specific areas, performing them in a more focused and efficient way, ultimately not only reducing the costs of your field inspection activities, but also increasing its success rate.

Some concrete examples of applying AI algorithms for feature recognition, related to building inspections, include the automatic detection of roofs with solar panels, or the automatic creation of a building inventory by detecting, and classifying, buildings with a specific roof type.

Increased Inspection Efficiency

Knowing where certain events take place, or knowing where certain objects are located, enables you to perform your inspection tasks in a focused, rather than random, way. This not only significantly increases the efficiency of your inspection tasks, but also reduces costs.

Superior Insights into Objects and Land Use

Increase your knowledge on the characteristics of your objects and land use by extracting information from aerial imagery to get insights into the amount of objects that are present in an area, the state of these objects, or the land use in a certain area.

Our Feature Recognition Services

Avineon not only helps you find the right AI algorithm tailored to your specific needs, but we also implement it by training it on so-called “training data.” This trained algorithm is then further improved by extensive testing on so-called “test data”. As a final step, the AI algorithm is then used on the actual data of interest.

We thus not only provide you with a working algorithm, but also, if desired, run it on your actual data of interest. This can be done only once, or in a periodic way. Repeatedly applying the algorithm will only improve the quality of its output. The results are provided both as a report and as actual data – which not only include the recognition of features, but also the actual creation of geospatial data from the detection.

Design

Avineon helps you in your quest to find the most suited algorithm for your needs and requirements.

Finding the Best Fitted Algorithm

A lot different algorithms exist, and they all fit different purposes. Finding the best fitting algorithm for your specific need and requirement is of key importance to ensure success.

Avineon has a broad knowledge of existing algorithms and can help you find the right one for you – we search the haystack and find your needle!

Finding the Best Fitted Algorithm

A lot different algorithms exist, and they all fit different purposes. Finding the best fitting algorithm for your specific need and requirement is of key importance to ensure success.

Avineon has a broad knowledge of existing algorithms and can help you find the right one for you – we search the haystack and find your needle!

Implement

Avineon trains, and tests, your algorithm. We can then, depending on your preference, either run the algorithm on your actual data of interest, or provide you with the working algorithm so can you run it yourself. In case of the latter, we can help set up the required IT environment to get you started.

Implementation of the Algorithm

Avineon has the capability to train, and test, your algorithm until it achieves the desired results.

We train your algorithm on so-called “training data” and then further improve it by extensive testing on so-called “test data”.

Feature Recognition Execution and Correction

Avineon runs your algorithm on your actual data of interest – meaning the actual data that need to be investigated. We thus not only provide you with a working algorithm, but also with the results of its implementation.

These results are provided to you as both a report and actual data. The data not only include the recognition of features, but also the actual creation of geospatial data from the detection.

Set-Up of the IT Environment

Alternatively, you can decide to run the algorithm yourself.

In that case, Avineon can help you set up the required IT environment to do so – we set up the IT environment to get you started.

Labelling of data

Avineon has built the software and tools to segment your street view imagery by recognizing object boundaries to high a level of accuracy.

 

Objects can be classified and labelled to a high degree of accuracy. 2D or 3D bounding boxes can created around specific type of objects like cars, bicycles, humans, vegetation, buildings etc on a street view imagery with minimal manual interventions, by applying our trained datasets and custom-algorithms on your datasets.

Implementation of the Algorithm

Avineon has the capability to train, and test, your algorithm until it achieves the desired results.

We train your algorithm on so-called “training data” and then further improve it by extensive testing on so-called “test data”.

Feature Recognition Execution and Correction

Avineon runs your algorithm on your actual data of interest – meaning the actual data that need to be investigated. We thus not only provide you with a working algorithm, but also with the results of its implementation.

These results are provided to you as both a report and actual data. The data not only include the recognition of features, but also the actual creation of geospatial data from the detection.

Set-Up of the IT Environment

Alternatively, you can decide to run the algorithm yourself.

In that case, Avineon can help you set up the required IT environment to do so – we set up the IT environment to get you started.

Labelling of data

Avineon has built the software and tools to segment your street view imagery by recognizing object boundaries to high a level of accuracy.

 

Objects can be classified and labelled to a high degree of accuracy. 2D or 3D bounding boxes can created around specific type of objects like cars, bicycles, humans, vegetation, buildings etc on a street view imagery with minimal manual interventions, by applying our trained datasets and custom-algorithms on your datasets.

Sustain and Grow

We immediately, and automatically, use the results of the feature recognition execution and correction for further geospatial analysis. The algorithm will be improved over time by running it repeatedly over newly acquired data.

Geospatial Analysis

In addition, we have the needed geospatial knowledge to conduct further geospatial analysis. By automatically combining the results of the algorithm with other sources of (geospatial) data, a deeper level of knowledge and insight can be obtained.

The results of this analysis will also be provided to you.

Periodic Running

If desired, Avineon can run your algorithm in a periodic way. Repeatedly running the algorithm over newly acquired data will increase the quality of its outputs.

In addition, the periodic running of your algorithm allows you to check for recurring events or track progress and situational changes.

Geospatial Analysis

In addition, we have the needed geospatial knowledge to conduct further geospatial analysis. By automatically combining the results of the algorithm with other sources of (geospatial) data, a deeper level of knowledge and insight can be obtained.

The results of this analysis will also be provided to you.

Periodic Running

If desired, Avineon can run your algorithm in a periodic way. Repeatedly running the algorithm over newly acquired data will increase the quality of its outputs.

In addition, the periodic running of your algorithm allows you to check for recurring events or track progress and situational changes.

Why Avineon

Avineon offers everything you need for automated feature recognition. Our “all-in” solution covers all aspects of the value chain – we not only have the experience to choose the most suited algorithm for your requirements, but we also have the platform needed to train, test, and run the algorithm.

 

By making use of our geospatial knowledge, we can, in addition, combine the results of the algorithm automatically with other sources of data for further geospatial analysis. If you prefer to run the algorithm yourself, Avineon has you covered as well – we set up the needed IT environment to get you started.

Fast Results Through Experience

Avineon has developed a ready-to-use platform for feature recognition, which means that we can quickly train, test, and run your best fitted algorithm.

Both Recognition
and Creation

We not only automatically recognize features on imagery, but also provide feature statistics, or even convert features into geospatial objects.

Immediate Usage for Geospatial Analysis

Thanks to our geospatial expertise, we can integrate the results of the feature recognition with other (geo) data, immediately using your feature recognition results for further geospatial analysis.

Delivery Models

At Avineon, we have the delivery model tailored to your needs:

 

> Delivery of the trained and tested algorithm: After selecting the most suited algorithm, and implementing this algorithm, we hand it over to you so you can run it in your environment as much as you need.

 

> Our “all-in” feature recognition service: After selecting the best fitted algorithm, and implementing this algorithm, we periodically run it for you.

 

Improving Your Business

In this guide (PDF) we explain how Avineon can help to improve your business by visualizing your systems, assets, and data.
 

Interested? Let's talk!