Telecom companies often have to make critical decisions about the height of the mast required based on whether the client provided latitude-longitude coordinates are on ground or on roof-top. The mast height directly impacts the cost of the project and is a critical consideration for telecom companies. To address this issue, we have developed our AI model - 3D Spotter that is an advanced application to tell whether the given geo-coordinates are on rooftop or ground. It uses satellite images as input data as opposed to cluster maps thereby improving accuracy of prediction.
How it works
The input would be the given geo-coordinates and satellite images of good resolution
The satellite images are pre-processed using advanced Computer Vision (CV) techniques
This enhances the quality around the coordinates thereby improving overall resolution
The satellite image is now passed through a customized variant of Resnet AI CV model pre-trained to detect rooftop
As a final step the processed image from the first AI model is passed through a second AI model
This determines whether the given coordinates are on ground or rooftop, the final output of the model
Use of advanced Computer Vision architecture that can achieve excellent accuracy
Minimal use of data for training the second AI model
Cost effective method to perform rooftop/ ground detection
Custom architecture that can do the computation much faster than conventional methods
3D Spotter is a specialised AI model for Telecommunications sector and the primary application of this is to determine the location of the geo coordinates provided by telecom customers for antenna placement lie on ground or on rooftop. This helps telecom companies determine feasibility and provide an accurate cost, feasibility and effort estimate to their customers.