DATOS holds AI4EO Training with PSA

26 Mar 2020 3:07 PM

The Remote Sensing and Data Science: DATOS Help Desk hosted another week-long Artificial Intelligence for Earth Observation (AI4EO) Training with the [technical] representatives from eight (8) divisions of Philippine Statistics Authority. 

PSA Staff and researchers pose in front of the satellite dish covered in radome installed in DOST-ASTI Building

Held last 2 to 6 March 2020, and attended by 27 participants, the AI4EO is set to help PSA with their 2022 Census on Agriculture and Fisheries (CAF).  

Just like the previous AI4EO workshops, the training started with an introduction of the DATOS Project, along with the different science and technology infrastructures of DOST-ASTI. Previous speakers from the COARE and PREGINET also discussed their services, while the PEDRO Center was presented by Joshua Frankie Rayo, Science Research Specialist from the ground receiving station team. 

Mr. Bani Lara, Sr. SRS, discussing PREGINET and their existing efforts nationwide
Ms. Glenda Baldonado, Information Systems Analyst II, explaining how researchers, members from the academe and gov't can access the COARE services for their R&D efforts
Joshua Frankie Rayo, SRS II of the PEDRO Center discussing the earth observation technologies of DOST-ASTI

Representatives from PSA’s eight divisions were also able to share their existing capabilities and initiatives that could help in better understanding the needs of PSA with the use of earth observation technologies.  

 

The succeeding days were set for lectures about the basic Remote Sensing technologies used by the DATOS Project that PSA could also adopt for their CAF. Guidelines on the development of training data for neural networks as well as Artificial Intelligence concepts and theories were also included in the lecture. 

On the third day, a parallel session was conducted for coding and digitization exercises. The researchers and technical staff of PSA were grouped into five (5). Each group had a programmer who went to learn and build their own artificial network, while the rest generated their training data for different features (mango trees, coconut trees, fishponds, and fish pens).  

  

The last two days were meant for the presentation of initial output of each group, retraining their models to achieve better result, and presentation of their final output. All of which are deemed success.  

On the last day, representatives from PSA expressed their gratitude and other concerns about the week-long training. Said that the transfer of learning was achieved, the PSA representatives expressed that they are looking forward to the next steps of improvement using remote sensing, AI, and EO data for PSA projects especially for their census for crops, aquaculture, and fisheries.  

 With the equally aligned ideals for nationwide implementation of earth observation techniques and technologies, the DATOS Project assured PSA that this collaboration will not end with the training. Continuous consultations are always welcome as long as the initiatives are aligned with the use of DOST-ASTI infrastructures for the 2022 Census.  

 

DOST-ASTI and PSA will work hand-in-hand for the improvement of both agencies’ mandate for nationwide application. Indeed, it is always Science for the People.