https://github.com/alexeyev/hogweed-ground-level-view
A dataset for semantic segmentation of Sosnowsky's hogweed in the ground-level view photos taken in St. Petersburg, Malaya Vishera, Pushkin, etc.
Science Score: 23.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
○codemeta.json file
-
○.zenodo.json file
-
✓DOI references
Found 3 DOI reference(s) in README -
✓Academic publication links
Links to: ieee.org, zenodo.org -
○Committers with academic emails
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (12.4%) to scientific vocabulary
Keywords
Repository
A dataset for semantic segmentation of Sosnowsky's hogweed in the ground-level view photos taken in St. Petersburg, Malaya Vishera, Pushkin, etc.
Basic Info
Statistics
- Stars: 3
- Watchers: 2
- Forks: 1
- Open Issues: 0
- Releases: 0
Topics
Metadata Files
README.md
Detecting Hogweed on the Ground-Level View Photographs: Dataset
Hogweed (Heracleum) is a herbs genus that features many invasive species such as giant hogweed or Sosnowsky's hogweed. This invasive species are particularly notorious due to the high content of phototoxic compounds, so that any contact with a plant may result in an intense skin burn.
Invasion of the Sosnowsky's hogweed [lang:RU] in particular is major trouble in Central Russia, and by 2021 resolving the problem requires massive intervention. Agtech drones spraying herbicides are already used to eradicate the Sosnowsky's hogweed, and accompanying real-time detection algorithms for UAVs are being developed (e.g. see this paper and the related dataset repository).
We propose a dataset for detecting Sosnowsky's hogweed using the ground-level view as if we're looking through the camera of an autonomous unmanned ground vehicle patrolling the hogweed-endangered area (e.g. a week after mowing or poisoning). It is not 100% clear whether this dataset can or should be used for training actual robotic vision algorithms or synthetic datasets construction. However, plant detection in the natural environment is quite a challenge, which makes such annotated images collections suitable for competitions and/or ML homeworks. This is a grassroot (pun intended) initiative without any external funding.
Data
Photographic images for the directory prepared_data/images/ (CC-BY-4.0) can be downloaded from Zenodo: 5233380.
444 (311/133) photos are taken in different locations in Russia using a Samsung Galaxy A31 camera. The images are annotated using https://supervise.ly/ (CE).
A more detailed description of the data collection strategy and the dataset in general will be released during autumn. Test set annotations will be released after the end of the competition.
Format
The annotations are provided in COCO format. To inspect the annotations manually, please see
the Jupyter notebook COCO-formatted-annotations-viewer.ipynb adapted from
the original Gist
shared by akTwelve.
Classification
To train a classifier,
- run a
get_data.shscript, - check out the Dataset object provided in
dataset.pyif you are planning to use PyTorch, - consider using a baseline implemented in
prepared_pipeline_for_transfer.py-- based on a fine-tunedResNet18model prepared by Dustin Franklin @dusty-nv. The training process is described in the tutorial. The model is available for downloading. All rights are reserved by NVIDIA.
How to cite
We would appreciate if you cite this dataset as
@dataset{alekseev_anton_2021_5233380,
author = {Alekseev, Anton},
title = {{Detecting Hogweed on the Ground-Level View Photographs: Dataset}},
month = aug,
year = 2021,
publisher = {Zenodo},
version = {0.1},
doi = {10.5281/zenodo.5233380},
url = {https://doi.org/10.5281/zenodo.5233380}
}
Acknowledgements
I would like to thank Aleksey Artamonov, Andrey Savchenko and Mikhail Evtikhiev for various consultations and proofreading.
Other materials
- A monster that devours Russia [YouTube video]
- Different species, similar threat: Giant Hogweed - The UK's Most Dangerous & Toxic Plant [YouTube video, possibly disturbing content]
Owner
- Name: Anton Alekseev
- Login: alexeyev
- Kind: user
- Website: https://ai.pdmi.ras.ru/
- Repositories: 52
- Profile: https://github.com/alexeyev
GitHub Events
Total
Last Year
Committers
Last synced: about 2 years ago
Top Committers
| Name | Commits | |
|---|---|---|
| alexeyev | a****v@g****m | 44 |
Issues and Pull Requests
Last synced: 12 months ago
All Time
- Total issues: 0
- Total pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Total issue authors: 0
- Total pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Past Year
- Issues: 0
- Pull requests: 0
- Average time to close issues: N/A
- Average time to close pull requests: N/A
- Issue authors: 0
- Pull request authors: 0
- Average comments per issue: 0
- Average comments per pull request: 0
- Merged pull requests: 0
- Bot issues: 0
- Bot pull requests: 0
Top Authors
Issue Authors
Pull Request Authors
Top Labels
Issue Labels
Pull Request Labels
Dependencies
- pytorch/pytorch 1.9.0-cuda10.2-cudnn7-runtime build
- pandas >=1.3.1
- scikit-learn >=0.24.2
- torch >=1.9.0
- torchvision >=0.10.0
- zenodo-get >=1.3.2