Internship: Object Detection in Georeferenced Streaming Video
In recent years there have been more and more interest in using machine learning for tasks like image recognition and object detection. During this period, with developments like the open source TensorFlow library, it has also become much easier to make use of these methods in software projects.
Another recent development is the widespread availability of drones that can deliver high resolution, geographically localized streaming video.
The purpose of the internship is to investigate whether existing and publically available object detection models can be integrated into the functionality for handling georeferenced video in Carmenta’s geospatial software.
The object detection models should be applied to recognize objects such as moving vehicles in a streamed georeferenced video from a drone. As a means of evaluation, the different vehicles should be georeferenced and visualized on a map together with the video itself.
A possible next step is to apply geospatial data fusion algorithms to correlate the detected vehicles with information from other sources such as radar tracks.
This internship is suitable for students interested in working with modern machine learning and geospatial analysis frameworks.
The internship is to be performed at the Carmenta office in Paris with support from the Carmenta R&D team in Sweden. When starting up your work you will be invited to Sweden to meet the team and learn more about the company.