4_Aerial Models

💾 FloodNet

• FloodNet is a dataset of very high resolution (cm) Uncrewed Aerial Vehicle (UAV) imagery, labeled according to the following classes:

1. Background
2. Building-flooded
3. Building-non-flooded
4. Road-flooded
5. Road-non-flooded
6. Water
7. Tree
8. Vehicle
9. Pool
10. Grass

• FloodNet dataset
• challenge
• paper

Example labeled imagery

Model for 768x512 imagery

• Zenodo release: https://zenodo.org/record/7566810#.Y-FljxzMLRY
• Buscombe, Daniel. (2023). Doodleverse/Segmentation Zoo/Seg2Map Res-UNet models for FloodNet/10-class segmentation of RGB 768x512 UAV images (v1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7566810
• segmentation zoo model name: floodnet_10class_7566810

Training curve for the best model:

Some example validation outputs:

Model for 1024x768 imagery

• Zenodo release: https://zenodo.org/record/7566797#.Y-FlkRzMLRY
• Buscombe, Daniel. (2023). Doodleverse/Segmentation Zoo/Seg2Map Res-UNet models for FloodNet/10-class segmentation of RGB 1024x768 UAV images (v1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7566797
• segmentation zoo model name: floodnet_10class_7566797

Training curve for the best model:

Some example validation outputs:

💾 NOAA-ERI

• NOAA-ERI is a dataset of high resolution (cm - m) oblique aerial imagery, labeled according to the following classes:

1. Water
2. Sediment
3. Vegetation
4. Development

• NOAA dataset
• Goldstein, Evan B., Buscombe, Daniel, Budavi, Priyanka, Favela, Jaycee, Fitzpatrick, Sharon, Gabbula, Sai Ram Ajay Krishna, Ku, Venus, Lazarus, Eli D., McCune, Ryan, Shah, Manish, Sigdel, Rajesh, & Tagner, Steven. (2022). Segmentation Labels for Emergency Response Imagery from Hurricane Barry, Delta, Dorian, Florence, Isaias, Laura, Michael, Sally, Zeta, and Tropical Storm Gordon (Version v1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7268083

Example labeled imagery

Model for 768x512 imagery

• Zenodo release: https://zenodo.org/record/7631354#.Y_fWsIDMLRY
• Goldstein, Evan B. (2023). Doodleverse/Segmentation Zoo Res-UNet model for NOAA ERI/4-class segmentation of RGB 512x512 images (1.1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7631354
• segmentation zoo model name: noaa_4class_7631354

Training curve for the best model:

💾 Watermasking

• A dataset of high resolution (cm - m) oblique aerial imagery, labeled according to the following classes:

1. Water
2. Other

• dataset
• dataset citation: Goldstein, Evan B., Buscombe, Daniel, Budavi, Priyanka, Favela, Jaycee, Fitzpatrick, Sharon, Gabbula, Sai Ram Ajay Krishna, Ku, Venus, Lazarus, Eli D., McCune, Ryan, Shah, Manish, Sigdel, Rajesh, & Tagner, Steven. (2022). Segmentation Labels for Emergency Response Imagery from Hurricane Barry, Delta, Dorian, Florence, Isaias, Laura, Michael, Sally, Zeta, and Tropical Storm Gordon (Version v1) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.726808

Example labeled imagery

Model for 1024x768 NOAA ERI imagery

• Zenodo release: https://zenodo.org/record/7604083#.Y_rrpIDMLRY
• model citation: Buscombe, Daniel. (2023). Doodleverse/Segmentation Zoo Res-UNet models for Aerial/NOAA ERI/2-class (water, nowater) segmentation of RGB 1024x768 high-res. images (v1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7604083
• segmentation zoo model name: noaa_2class_7604083

Training curve for the best model:

Some example validation outputs: