Assessment of Forest Loss Following Snow and Ice Storms Using the LiDAR Forest Structure Change Index

Plant Phenomics, Год журнала: 2025, Номер unknown, С. 100057 - 100057

Опубликована: Май 1, 2025

Язык: Английский

---

Automatic Detection of Phytophthora pluvialis Outbreaks in Radiata Pine Plantations Using Multi-Scene, Multi-Temporal Satellite Imagery

Remote Sensing, Год журнала: 2024, Номер 16(2), С. 338 - 338

Опубликована: Янв. 15, 2024

This study demonstrates a framework for using high-resolution satellite imagery to automatically map and monitor outbreaks of red needle cast (Phytophthora pluvialis) in planted pine forests. methodology was tested on five WorldView scenes collected over two sites the Gisborne Region New Zealand’s North Island. All were acquired September: four yearly (2018–2020 2022) Wharerata, while one more obtained 2019 Tauwhareparae. Training areas selected each scene manual delineation combined with pixel-level thresholding rules based band reflectance values vegetation indices (selected empirically) produce ‘pure’ training pixels different classes. A leave-one-scene-out, pixel-based random forest classification approach then used classify all images into (i) healthy forest, (ii) unhealthy or (iii) background. The overall accuracy models internal validation dataset ranged between 92.1% 93.6%. Overall accuracies calculated left-out 76.3% 91.1% (mean 83.8%), user’s producer’s across three classes 60.2–99.0% (71.4–91.8% forest) 54.4–100% (71.9–97.2% forest), respectively. work possibility classifier trained set new completely independent scenes. paves way scalable largely autonomous health monitoring system annual acquisitions at time peak disease expression, greatly reducing need interpretation delineation.

Язык: Английский

---

Space to depth convolution bundled with coordinate attention for detecting surface defects

Signal Image and Video Processing, Год журнала: 2024, Номер 18(5), С. 4861 - 4874

Опубликована: Апрель 1, 2024

Язык: Английский

---

Evaluating a Novel Approach to Detect the Vertical Structure of Insect Damage in Trees Using Multispectral and Three-Dimensional Data from Drone Imagery in the Northern Rocky Mountains, USA

Remote Sensing, Год журнала: 2024, Номер 16(8), С. 1365 - 1365

Опубликована: Апрель 12, 2024

Remote sensing is a well-established tool for detecting forest disturbances. The increased availability of uncrewed aerial systems (drones) and advances in computer algorithms have prompted numerous studies insects using drones. To date, most used height information from three-dimensional (3D) point clouds to segment individual trees two-dimensional multispectral images identify tree damage. Here, we describe novel approach classifying the reflectances assigned 3D cloud into damaged healthy classes, retaining assessment vertical distribution damage within tree. Drone were acquired 27-ha study area Northern Rocky Mountains that experienced recent then processed produce cloud. Using data points on (based depth maps images), random (RF) classification model was developed, which had an overall accuracy (OA) 98.6%, when applied across area, it classified 77.0% with probabilities greater than 75.0%. Based segmented trees, developed evaluated separate trees. For identified severity each based percentages red gray top-kill length continuous treetop. Healthy separated high (OA: 93.5%). remaining different severities moderate 70.1%), consistent accuracies reported similar studies. A subsequent algorithm 91.8%). as (78.3%), exhibited some amount (78.9%). Aggregating tree-level metrics 30 m grid cells revealed several hot spots severe illustrating potential this methodology integrate products space-based remote platforms such Landsat. Our results demonstrate utility drone-collected monitoring structure diseases.

Язык: Английский

---

Automatic detection of faults in industrial production of sandwich panels using Deep Learning techniques

Logic Journal of IGPL, Год журнала: 2024, Номер unknown

Опубликована: Май 2, 2024

Abstract The use of technologies like artificial intelligence can drive productivity growth, efficiency and innovation. goal this study is to develop an anomaly detection method for locating flaws on the surface sandwich panels using YOLOv5. proposed algorithm extracts information locally from image through a prediction system that creates bounding boxes determines whether panel contains flaws. It attempts reject or accept product based quality levels specified in standard. To evaluate method, comparison was made with damage convolutional neural network methods thresholding. findings show which object detector, more accurate than alternatives. characteristics model, according standard limit allowable manufacturing obtain product, also enable improve industrial standards producing while increasing speed.

Язык: Английский

---