Counterpart training in radar (SAR) satellite data analysis and machine learning (AI) was conducted by JAXA and AIST from September 12 to 16.
Originally, the training was planned to be conducted in Japan, but due to the uncertainty of the timing of the trip due to COVID-19, the experts traveled from Japan for the training this time. In the SAR analysis training conducted by JAXA, the participants examined the patterns of deforestation in the Amazon and how SAR data changes during such changes in order to improve the accuracy of early detection of forest change. In the AI programming training conducted by AIST, participants learned about machine learning methods using Python. Although the mechanics and mathematical formulas of deep learning seemed difficult, the incorporation of concrete examples helped the participants to better understand the subject. While this training was useful for the Brazilian side to learn the technology, it was a good opportunity for the Japanese side to understand the situation of illegal deforestation in Brazil through the discussions during the training. From now on, we will continue to share information online on a regular basis and collaborate with each other to promote project activities. In addition, on the 15th, a seminar was held for not only the training participants but also the staff in IBAMA to let them know about the project, JAXA, and AIST’s activities, which was attended by about 50 people, including online participants.