Geo4i has developed geospatial analysis support solutions and a platform for the exploitation of images and mass data.


GeoSpace is a platform dedicated to imagery and geospatial analysis within Big Data context
Download the leaflet here


Help4i is a tools dedicated to equipment identification based on visual criteria

Donwload the leaflet here


Artificial Intelligence

Artificial Intelligence has become a major focus in the news.
Applied to space imaging in the field of image intelligence, results are still few and applications are more developed for marketing purposes than operational one (not industrialized solutions yet).
If we consider that AI gives good results, its advantages are to save time during the analysis (less documentary research), to capitalize the information, to possibly carry out processing on massive information (alert level settings on imagery which would not have been visualized).
In a context of statistical elements analysis, the requirement on the level of the detection rate is less constrained (for example “forgetting” few elements among several thousands will have unsignificative influence on the result).
However, if the AI process has not been carry out properly, the risks induced by blind trust in the tools could have a major impact in a military context for example.
Indeed, an error of analysis on an area of operation can have a significant strategic impact (AI is then misleading) or worse can cause human losses.
In this context, skilled knowledge is essential in setting up the base and learning processes: it is difficult to entrust these tasks to inexperienced “clickworkers”.
For Geo4i, which has done work in the field (see video), the artificial intelligence tools are, for the moment, only a help and does not replace the analyst who has a sensitivity, an apprehension and experience taking into account the context.
Even if the result seems good, it must be checked and considered more as an help than a truth.

Geo4i’s efforts are focused on a training dataset and a model according our workflow and the tools we have already developed.
Several iterations were necessary to improve the results. We are at the fourth.
In parallel, our development focuses on the development of a labeling tool for GeoSpace platform users.

To book a demo, contact us