| Full text | |
| Author(s): |
Casagrande, Luan
;
Hirata, R., Jr.
Total Authors: 2
|
| Document type: | Journal article |
| Source: | IGARSS 2024-2024 IEEE INTERNATIONAL GEOSCIENCE AND REMOTE SENSING SYMPOSIUM, IGARSS 2024; v. N/A, p. 5-pg., 2024-01-01. |
| Abstract | |
Riparian zones play a crucial role in water resources, wildlife, and communities. Governments have regulations to protect these areas, and quickly and accurately mapping vegetation near rivers to ascertain compliance with regulations is crucial. We propose the use of UAV data to calibrate a Sentinel-2 based model to predict class membership in riparian zones. In comparison to similar works, the proposed approach based on Convolutional Neural Network calibrated by a DeepLabV3+ is significantly better when evaluating the dominant class and has a higher potential to describe class membership for heterogeneous pixels. (AU) | |
| FAPESP's process: | 22/15304-4 - Learning context rich representations for computer vision |
| Grantee: | Nina Sumiko Tomita Hirata |
| Support Opportunities: | Research Projects - Thematic Grants |