In assessing the suitability of invented names, the Agency of Medicines verifies that the name cannot be confused with the name of another medicinal product and does not contain the name of the active substance. The Agency of Medicines performs approximately 1,200 assessments of invented names per year. On average, it takes a specialist around 7 minutes to assess 1 name, which adds up to at least 138 hours of working time over the year. Automated name verification would provide specialists with freed-up working time that they could dedicate to conducting substantive assessments of new applications. Automated name verification not only optimises the use of working time but also reduces the potential for human error, which is high in manual name verification.

The virtual assistant ITI, who can answer the typical questions of statistics consumers and reporting agents, assists customers on the website of Statistics Estonia. If ITI does not know the answer, it will forward the enquiry to customer support based on its content. The virtual assistant ITI knows the population of Estonia, the size of the wage gap, average wages and consumer price changes, and many other things. ITI also assists the data provider whose typical questions are related to activities in the electronic data submission portal and filling in questionnaires.

Veriff's biometric facial recognition technology developed in Estonia allows for the secure verification of the identity document and the person involved in the transaction. Remote authentication is not comparable to communicating or performing operations via Skype or Zoom. The new system has been years in the making, as the verification of the identity and the actual intention of the person involved in the notarial act is of critical importance. Enabling notarial acts to be performed remotely was not only necessary during the pandemic but also under normal circumstances where one party to a transaction is located far from Estonia and travelling here would be time-consuming, costly or otherwise challenging.

As part of the project, in collaboration with Estonian universities, a methodology was developed for crop identification in the context of Estonia. The study combined data from the Sentinel 1 and 2 satellites on soil type, daily average air temperature and precipitation and showed that crop classes can be identified with over 90% accuracy for potatoes, peas, broad beans, summer and winter rapeseed, field mustard, maize, rye, winter wheat and winter barley. It was not possible to identify with sufficient accuracy berry and fruit tree plantations, carrots and beetroot.
The developed methodology is planned to be used both in ARIB and for the benefit of Estonia as a whole � to generate objective and consistently high-quality agricultural statistics.

A Python-based framework that pre-processes X-Road logs, prepares and forwards usage reports to members, displays anomalies and anonymises and publishes log data as open data. The system can provide guidance to users based on anomalies and meets the requirements for open data in the public sector.

The Environment Agency and KEMIT have, together with a development partner, created a remote sensing information system for forests which enables records of forest resources to be kept in a geo-referenced manner, and forest information to be collected and shared in an operative manner. The created solution is semi-automatic, being partially based on machine learning algorithms, partially on the visual verification of remote sensing data, and partially on the software solutions of third parties. Remote sensing data are mainly used as primary data, but forest notifications, soil maps, and other such data (incl teaching and validation data) also act as inputs. The processing will create Estonia-wide map layers about more important survey characteristics, such as a raster map of tree species in woody flora (for identification based on the main tree species).

Ilme enables users to search for photographic images of both famous historical figures and their ancestors by their appearance. Ilme uses facial recognition software to find persons whose facial features most closely resemble the ones captured in the uploaded photo in the Fotis photo database of the National Archives of Estonia. To search, users simply have to upload a photo of their interest to the Ilme environment, after which the system will use facial recognition to identify and compare faces with more than half a million photos spanning from the 1840s onwards. Within 24 hours, the photo uploader receives a PDF file via email, containing the most similar matches from a pool of around three million faces.

The aim of the project was to create a mathematical machine learning model for forecasting traffic accidents, evaluate its accuracy and usability, and to create a prototype of the model.

A map application based on satellite-based ice products, in which the application of machine learning was also tested.

Estonia's first self-driving first/last kilometre vehicle was built in 2018 by the Tallinn University of Technology. It is a bus that operates without a traditional steering wheel, using sensors, radars and lidars instead. Over the years, the self-driving bus, named iseAuto, has undergone further developments and, in summer 2020, received permission from the Road Administration to operate in regular traffic. The first testing took place in the summer of 2020 on a fixed 2.4 km route in Ülemiste City, serving residents from 10:00 to 16:00. This has been followed by trials in Rae municipality and Tartu.

The Environment Agency and KEMIT have, together with a development partner, created a remote sensing information system for forests which enables records of forest resources to be kept in a geo-referenced manner, and forest information to be collected and shared in an operative manner. The created solution is semi-automatic, being partially based on machine learning algorithms, partially on the visual verification of remote sensing data, and partially on the software solutions of third parties. Remote sensing data are mainly used as primary data, but forest notifications, soil maps, and other such data (incl teaching and validation data) also act as inputs. The processing will create Estonia-wide map layers about more important survey characteristics, such as a vector map of felled areas (for identification of clear cutting > 0.5 ha).

Using drones to search for people