Object detection of the bornean orangutan nests using drone and YOLOv5

IAES International Journal of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 13(2), P. 1640 - 1640

Published: April 5, 2024

<p class="tm6">Object detection methods when applied to ecology and conservation can help identify monitor endangered species their habitats. Using drones for this purpose has become increasingly popular due ability cover large areas quickly efficiently. In study, we aim implement object using YOLOv5 detect orangutan nests in forests. To conduct our experiment, collect drone imagery under different conditions. We propose use the original model. The monitoring of conservationists critical habitats, population, design effective strategies. Additionally, reduce need on-the-ground surveys, which be time-consuming, expensive, logistically challenging. study proposes a model detecting forests YOLOv5. Our predicted 1,970 training images 414 labeled nests, with precision 0.973, recall 0.949, accuracy mean average (mAP)_0.5 is 0.969, mAP_0.5:0.95 0.630. finished 217 epochs 58 hours had high accuracy. 99.9% number nests.</p>

Language: Английский

---

Mapping large-scale pine wilt disease trees with a lightweight deep-learning model and very high-resolution UAV images

International Journal of Remote Sensing, Journal Year: 2024, Volume and Issue: 45(8), P. 2786 - 2807

Published: April 11, 2024

Pine wilt disease (PWD), caused by pine wood nematodes, has brought a great loss in ecology and economy all over the world. In China, forest health status is also significantly affected PWD since 1980, especially coniferous forests mixed regions. The spreads very fast can cause healthy tree to die within short time. An effective way protect other trees discover early. Using unmanned aerial vehicle (UAV) images help people quickly accurately, few automatical methods have been developed monitor including deep learning methods. Because of robust spatial-temporal transferability, become mainstream algorithms trees. As we know, training dataset most important material train model. However, there still lack segmentation so far. To fill this gap, paper, generated first open-sourced based on high-resolution UAV community conduct research conveniently. This 994 samples, each sample visible bands with 512 × pixels, spatial resolution 0.05 m. order an advanced model, designed lightweight deep-learning model for mobile devices or edge manuscript, named MobileSeg. main feature MobileSeg its decoupling, which uses re-parameterization technology improve performance. Finally, large-scale real-world scenario experiment was utilized validate performance MobileSeg, result indicated that achieved best compared recent models, proved effectiveness proposed dataset.

Language: Английский

---

Determination of the effective number of hunting animals based on geoformation data about hunting grounds

AIP conference proceedings, Journal Year: 2024, Volume and Issue: 3021, P. 080003 - 080003

Published: Jan. 1, 2024

Views Icon Article contents Figures & tables Video Audio Supplementary Data Peer Review Share Twitter Facebook Reddit LinkedIn Tools Reprints and Permissions Cite Search Site Citation Artеm Rada, Aleksandr Kuznetsov, Roman Zverev; Determination of the effective number hunting animals based on geoformation data about grounds. AIP Conf. Proc. 29 March 2024; 3021 (1): 080003. https://doi.org/10.1063/5.0193485 Download citation file: Ris (Zotero) Reference Manager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex toolbar search Dropdown Menu input auto suggest filter your All ContentAIP Publishing PortfolioAIP Conference Proceedings Advanced |Citation

Language: Английский

---

Regulation of Interaction between Hunters and Land Users: A Comparative Legal Study

Laws, Journal Year: 2023, Volume and Issue: 12(1), P. 14 - 14

Published: Jan. 30, 2023

Hunting is a complex type of nature management. In its process, objects the animal world and earth are used. Obviously, relationship between hunters other land users should be clearly regulated by legislation. The purpose this work was to identify common specific problems for different systems interaction owners assess possibility spreading existing experience solving faced hunting sector countries. Three main models (direct interaction, cooperation, division rights) considered. Each performs tasks has own degree efficiency. organization model adopted in country depends on specifics conditions which farm develops including economic, property, legal, social, state aspects. It established that availability best ensured within framework cooperation model, observation rights owners—within direct convenience management large territories wild habitats—within model. At same time, it incorrect single out all criteria or designate universally suitable conditions. farms Russia, described interactions not related potential as such, but lack understanding particular requires increased attention state. proposals aimed at improving practice developing applying relationships represented.

Language: Английский

---

Data Normalization Technique for Energy Efficient Drones: An Ensemble Learning Approach

2021 IEEE 12th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON), Journal Year: 2022, Volume and Issue: unknown, P. 0411 - 0416

Published: Oct. 12, 2022

Frequent wildfires in the western part of United States are affecting local economy, flora and fauna, air quality, health community. Over years, various techniques have been utilized for early detection wildfires, that include satellites, lookout towers, drones. Among techniques, drones gaining popularity due to recent advancements drone technology, multi-role adaptability, lower cost operation. However, be effective during disaster management monitoring, endurance longevity drone's flight time essential. A transistor-embedded photovoltaics (PV) panel-powered can enable such essential qualities required drone. Such a power source requires an efficient algorithm switching configuration PV panel them different lighting operating conditions. Machine learning classification as Random Forest activating shown effectiveness detecting presence shade. with larger number properties training, supervised ML result increased memory usage and, some cases, accuracy. In this paper, we propose novel normalization technique reduce train machine model. After applying technique, We observed performance model 90.1 % shade module, along 7.535 reduction usage.

Language: Английский

---