This project is a collaboration with the “Tour du Valat” association. It has received contributions from several master students at GT-Europe. The results presented here have been developed by Tom Bourbon. The project objective is to optimize the time taken by staff to read the tags placed on a bird species characteristic of the Camargue area in southern France, the white spoonbill (Platalea Leucorodia). The tags are used to identify individuals. Knowing the dated position of an individual is a valuable piece of data, used in particular for studying migratory flows.

Typical image captures by the automated monitoring system.

From a typical image as seen above, our objective is to first to detect the tags:

Automated tag detection

From the detected tags, a deep learning letter classifier can be trained:

Automated tag decoding

The resulting work has been integrated into a user interface dedicated to the ornithological studies.

User interface
Reported detection in the user interface