U-Net for crab image semantic segmentation with PyTorch : UAV imaging and deep learning approach can id entify Brachyura in tidal flats

Fuente: Zenodo
Salvato in:
Dettagli Bibliografici
Autori principali: Dongwoo Kim, Kwon Lee, Seungwoo Son
Natura: Recurso digital
Lingua:inglese
Pubblicazione: Zenodo 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866902186317840384
author Dongwoo Kim
Kwon Lee
Seungwoo Son
author_facet Dongwoo Kim
Kwon Lee
Seungwoo Son
contents <p>This dataset supports the study "Identifying Brachyura in tidal flats using UAV imaging and a deep-learning approach: A Republic of Korea west coast case study" (Scientific Reports). It contains the trained U-Net semantic segmentation model and associated scripts used to detect and classify four intertidal Brachyura species — Austruca lactea, Tubuca arcuata, Macrophthalmus japonicus, and Scopimera globosa, including sex-level discrimination for the first two species — from high-resolution UAV imagery collected at the Baramarae tidal flat, Taeanhaean National Park, Republic of Korea.<br>The repository includes: (1) best_epoch.pth, the trained U-Net model weights; (2) config.py, model configuration parameters; (3) dice_loss.py, the Dice loss function used during training; (4) evaluate.py, a script for computing pixel-level accuracy metrics; (5) predict.py, a script for running species predictions on new UAV images; and (6) test.py, a script for evaluating model performance on labelled test images. Raw UAV imagery and species labels are not included owing to the endangered status of A. lactea (nationally protected in the Republic of Korea). The code is based on the PyTorch U-Net implementation (https://github.com/milesial/Pytorch-UNet).</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_10589746
institution Zenodo
language eng
publishDate 2024
publisher Zenodo
record_format zenodo
spellingShingle U-Net for crab image semantic segmentation with PyTorch : UAV imaging and deep learning approach can id entify Brachyura in tidal flats
Dongwoo Kim
Kwon Lee
Seungwoo Son
Brachyura, Deep-learning algorithm, Ecosystem monitoring, Non-invasive monitoring, Tidal flats; UAV, U-Net model
<p>This dataset supports the study "Identifying Brachyura in tidal flats using UAV imaging and a deep-learning approach: A Republic of Korea west coast case study" (Scientific Reports). It contains the trained U-Net semantic segmentation model and associated scripts used to detect and classify four intertidal Brachyura species — Austruca lactea, Tubuca arcuata, Macrophthalmus japonicus, and Scopimera globosa, including sex-level discrimination for the first two species — from high-resolution UAV imagery collected at the Baramarae tidal flat, Taeanhaean National Park, Republic of Korea.<br>The repository includes: (1) best_epoch.pth, the trained U-Net model weights; (2) config.py, model configuration parameters; (3) dice_loss.py, the Dice loss function used during training; (4) evaluate.py, a script for computing pixel-level accuracy metrics; (5) predict.py, a script for running species predictions on new UAV images; and (6) test.py, a script for evaluating model performance on labelled test images. Raw UAV imagery and species labels are not included owing to the endangered status of A. lactea (nationally protected in the Republic of Korea). The code is based on the PyTorch U-Net implementation (https://github.com/milesial/Pytorch-UNet).</p>
title U-Net for crab image semantic segmentation with PyTorch : UAV imaging and deep learning approach can id entify Brachyura in tidal flats
topic Brachyura, Deep-learning algorithm, Ecosystem monitoring, Non-invasive monitoring, Tidal flats; UAV, U-Net model
url https://doi.org/10.5281/zenodo.10589746