5 May 2017
The Serbian Republic Geodetic Authority (RGA), under their World Bank Real Estate Management Project, is intending to commission a piece of work to extract building features over Serbia using very high resolution satellite imagery. The building extraction will to be done over the territory of Serbia, excluding AP Kosovo and Metohia (the full area is approximately 77, 550 km2, where artificial areas cover approximately 3 700 km2 – extracted from Generic Land Cover Map).
Satellite Applications Catapult has worked with the RGA to create this opportunity for organisations with expertise within this domain.
The critical output of the building feature extraction process is the identification and location of individual buildings. The priority for the full opportunity is rapid, cost-effective mapping of the majority of building over the territory of Serbia, thus trade-offs will be considered by RGA between accuracy of polygon feature edges and the speed and cost effectiveness of the process.
The current opportunity is for interested providers to demonstrate outputs of their feature extraction algorithms/methodology against a sample of 3 representative pilot areas in Serbia, to be submitted to RGA for evaluation and assessment.
The outputs from the pilot areas and any additional information indicating ROM pricing and timescales for delivering the full project as well as evidence of experience must be submitted by the closing date of this opportunity.
The pilot areas will be:
Upon signing a Data Use Statement, you will be sent the data for the pilot areas by members of the RGA.
RGA will engage with all organisations who submit a pilot study, and potentially request a formal proposals, by 2 May 2017.
The primary class for identification relates to Buildings.Thematic accuracy of building classification should not be less than 85%.
Advisable subclasses that are of interest are:
The satellite images to be used for the full study are orthorectified multispectral WorldView-2, WorldView-3 and GeoEye satellite images with Blue, Green, Red and Nir bands of 40 cm and 30 cm spatial resolution and of off nadir angle of the images ranging from 0o to 40o. The positional accuracy of orthorectified images is ± 2 x value of spatial resolution rmse and the images are georeferenced in the ETRS89/ UTM 34N projection.
You will be required to sign a Data Use Statement before accessing the imagery and upload it via the form below.
Hosting of the data after the pilot:
The data for the full study could be hosted in CEMS and accessed through a CEMS VM. Alternatively, the data could be provided to be processed elsewhere.
Respond to an Opportunity:
For further information, please contact: firstname.lastname@example.org