
Building an Image-classification Library for Deep-learning Algorithms
The purpose of BILDA is to improve the detection rate of the detection algorithm we use.
Until now, the detection rate has been too low for it to be possible to use only analyzes from the algorithm.
This means that a third party must manually verify data that the algorithm cannot approve on its own.
By creating an application where other people can verify the data from the algorithm by simply swiping on images,
it is expected that the detection rate can be improved with up to 95% accuracy,
which will drastically reduce resources spent on manually verifying the data.

THE FUTURE OF SURVEYING
Quantify your maintenance backlog automatically
OUR OPERATIONS
Aerial robotics, combined with powerful computer vision algorithms - Generates insightful data to asset owners and managers.








With artificial intelligence, deep learning algorithms and powerful sensors, ProXpect can map and populate your assets data within minutes.
Prioritize your maintenance proactively and ensure that vaulable assets is maintained in accordance with your preferred standards.
OUR PROCESS
STRUCTURE AND DATA HARVEST
THE COST EFFICIENT, SWIFT AND SAFE WAY
1. INTRO
A brief free consultation with the clients needs. Decide thresholds and goals for the survey, ensuring expectancy alignment.
2. SURVEYING
The area is surveyed using robotics and advanced calibrated sensors. The survey data is uploaded to your secure systems and verified.
3. INSPECTION
The data is inspected using advanced computer vision algorithms. The results is verified by our image analysts and engineers for quality assurance.
4. DELIVERY
The data is delivered digitally, either directly in to the clients maintenance software or as standard open formats.
THE SOLUTION



























CLIENTS




