YOLOv4-Lite–Based Urban Plantation Tree Detection and Positioning With High-Resolution Remote Sensing Imagery

Automatic tree identification and position using high-resolution remote sensing images are critical for ecological garden planning, management, and large-scale environmental quality detection. However, existing single-tree detection methods have a high rate of misdetection in forests not only due to the similarity of background and crown colors but also because light and shadow caused abnormal crown shapes, resulting in a high rate of misdetections and missed detection. This article uses urban plantations as the primary research sample. In conjunction with the most recent deep learning method for object detection, a single-tree detection method based on the lite fourth edition of you only look once (YOLOv4-Lite) was proposed. YOLOv4’s object detection framework has been simplified, and the MobileNetv3 convolutional neural network is used as the primary feature extractor to reduce the number of parameters. Data enhancement is performed for categories with fewer single-tree samples, and the loss function is optimized using focal loss. The YOLOv4-Lite method is used to detect single trees on campus, in an orchard, and an economic plantation. Not only is the YOLOv4-Lite method compared to traditional methods such as the local maximum value method and the watershed method, where it outperforms them by nearly 46.1%, but also to novel methods such as the Chan-Vese model and the template matching method, where it outperforms them by nearly 26.4%. The experimental results for single-tree detection demonstrate that the YOLOv4-Lite method improves accuracy and robustness by nearly 36.2%. Our work establishes a reference for the application of YOLOv4-Lite in additional agricultural and plantation products.

Comparison of Classical Methods and Mask R-CNN for Automatic Tree Detection and Mapping Using UAV Imagery

Detecting and mapping individual trees accurately and automatically from remote sensing images is of great significance for precision forest management. Many algorithms, including classical methods and deep learning techniques, have been developed and applied for tree crown detection from remote sensing images. However, few studies have evaluated the accuracy of different individual tree detection (ITD) algorithms and their data and processing requirements. This study explored the accuracy of ITD using local maxima (LM) algorithm, marker-controlled watershed segmentation (MCWS), and Mask Region-based Convolutional Neural Networks (Mask R-CNN) in a young plantation forest with different test images. Manually delineated tree crowns from UAV imagery were used for accuracy assessment of the three methods, followed by an evaluation of the data processing and application requirements for three methods to detect individual trees. Overall, Mask R-CNN can best use the information in multi-band input images for detecting individual trees. The results showed that the Mask R-CNN model with the multi-band combination produced higher accuracy than the model with a single-band image, and the RGB band combination achieved the highest accuracy for ITD (F1 score = 94.68%). Moreover, the Mask R-CNN models with multi-band images are capable of providing higher accuracies for ITD than the LM and MCWS algorithms. The LM algorithm and MCWS algorithm also achieved promising accuracies for ITD when the canopy height model (CHM) was used as the test image (F1 score = 87.86% for LM algorithm, F1 score = 85.92% for MCWS algorithm). The LM and MCWS algorithms are easy to use and lower computer computational requirements, but they are unable to identify tree species and are limited by algorithm parameters, which need to be adjusted for each classification. It is highlighted that the application of deep learning with its end-to-end-learning approach is very efficient and capable of deriving the information from multi-layer images, but an additional training set is needed for model training, robust computer resources are required, and a large number of accurate training samples are necessary. This study provides valuable information for forestry practitioners to select an optimal approach for detecting individual trees.

UAV-Based LiDAR Scanning for Individual Tree Detection and Height Measurement in Young Forest Permanent Trials

The installation of research or permanent plots is a very common task in growth and forest yield research. At young ages, tree height is the most commonly measured variable, so the location of individuals is necessary when repeated measures are taken and if spatial analysis is required. Identifying the coordinates of individual trees and re-measuring the height of all trees is difficult and particularly costly (in time and money). The data used comes from three Pinus pinaster Ait. and three Pinus radiata D. Don plantations of 0.8 ha, with an age ranging between 2 and 5 years and mean heights between 1 and 5 m. Five individual tree detection (ITD) methods are evaluated, based on the Canopy Height Model (CHM), where the height of each tree is identified, and its crown is segmented. Three CHM resolutions are used for each method. All algorithms used for individual tree detection (ITD) tend to underestimate the number of trees. The best results are obtained with the R package, ForestTools and rLiDAR. The best CHM resolution for identifying trees was always 10 cm. We did not detect any differences in the relative error (RE) between Pinus pinaster and Pinus radiata. We found a pattern in the ITD depending on the height of the trees to be detected: the accuracy is lower when detecting trees less than 1 m high than when detecting larger trees (RE close to 12% versus 1% for taller trees). Regarding the estimation of tree height, we can conclude that the use of the CHM to estimate height tends to underestimate its value, while the use of the point cloud presents practically unbiased results. The stakeout of forestry research plots and the re-measurement of individual tree heights is an operation that can be performed by UAV-based LiDAR scanning sensors. The individual geolocation of each tree and the measurement of heights versus pole and/or hypsometer measurement is highly accurate and cost-effective, especially when tree height reaches 1–1.5 m.

Accuracy of a LiDAR-Based Individual Tree Detection and Attribute Measurement Algorithm Developed to Inform Forest Products Supply Chain and Resource Management

Individual Tree Detection (ITD) algorithms that use Airborne Laser Scanning (ALS) data can provide accurate tree locations and measurements of tree-level attributes that are required for stand-to-landscape scale forest inventory and supply chain management. While numerous ITD algorithms exist, few have been assessed for accuracy in stands with complex forest structure and composition, limiting their utility for operational application. In this study, we conduct a preliminary assessment of the ability of the ForestView® algorithm created by Northwest Management Incorporated to detect individual trees, classify tree species, live/dead status, canopy position, and estimate height and diameter at breast height (DBH) in a mixed coniferous forest with an average tree density of 543 (s.d. ±387) trees/hectare. ITD accuracy was high in stands with lower canopy cover (recall: 0.67, precision: 0.8) and lower in stands with higher canopy cover (recall: 0.36, precision: 0.67), mainly owing to omission of suppressed trees that were not detected under the dominant tree canopy. Tree species that were well-represented within the study area had high classification accuracies (producer’s/user’s accuracies > ~60%). The similarity between the ALS estimated and observed tree attributes was high, with no statistical difference in the ALS estimated height and DBH distributions and the field observed height and DBH distributions. RMSEs for tree-level height and DBH were 0.69 m and 7.2 cm, respectively. Overall, this algorithm appears comparable to other ITD and measurement algorithms, but quantitative analyses using benchmark datasets in other forest types and cross-comparisons with other ITD algorithms are needed.

Refining and evaluating a Horvitz– Thompson-like stand density estimator in individual tree detection based on airborne laser scanning

Horvitz--Thompson-like stand density estimation is a method for estimating the stand density from tree crown objects extracted from airborne laser scanning data through individual tree detection. The estimator is based on stochastic geometry and mathematical morphology of the (planar) set formed by the detected tree crowns. This set is used to approximate the detection probabilities of trees. These probabilities are then used to calculate the estimate. The method includes a tuning parameter, which needs to be known to apply the method. We present a refinement of the method to allow more general detection conditions than the previous papers and present and discuss the methods for estimating the tuning parameter of the estimator using a functional $k$-nearest neighbors method. We test the model fitting and prediction in two spatially separate data sets and examine the plot-level accuracy of estimation. The estimator produced a $13$\% lower RMSE than the benchmark method in an external validation data set. We also analyze the effects of similarity and dissimilarity of training and validation data to the results.

Single Shot MultiBox Detector for Urban Plantation Single Tree Detection and Location With High-Resolution Remote Sensing Imagery

Using high-resolution remote sensing images to automatically identify individual trees is of great significance to forestry ecological environment monitoring. Urban plantation has realistic demands for single tree management such as catkin pollution, maintenance of famous trees, landscape construction, and park management. At present, there are problems of missed detection and error detection in dense plantations and complex background plantations. This paper proposes a single tree detection method based on single shot multibox detector (SSD). Optimal SSD is obtained by adjusting feature layers, optimizing the aspect ratio of a preset box, reducing parameters and so on. The optimal SSD is applied to single tree detection and location in campuses, orchards, and economic plantations. The average accuracy based on SSD is 96.0, 92.9, and 97.6% in campus green trees, lychee plantations, and palm plantations, respectively. It is 11.3 and 37.5% higher than the latest template matching method and chan-vese (CV) model method, and is 43.1 and 54.2% higher than the traditional watershed method and local maximum method. Experimental results show that SSD has a strong potential and application advantage. This research has reference significance for the application of an object detection framework based on deep learning in agriculture and forestry.

Optimizing aerial imagery collection and processing parameters for drone-based individual tree mapping in structurally complex conifer forests

Recent advances in remotely piloted aerial system (“drone”) and imagery processing technologies have enabled individual tree mapping in forest stands across broad areas with low-cost equipment and minimal ground-based data collection. One such method, “structure from motion” (SfM), involves collecting many partially overlapping aerial photos over a focal area and using photogrammetric analysis to infer 3D structure and detect individual trees. SfM-based forest mapping involves myriad decisions surrounding the selection of methods and parameters for imagery acquisition and processing, but no studies have comprehensively and quantitatively evaluated the influence of these parameters on the accuracy of the resulting tree maps.We collected and processed drone imagery of a moderate-density, structurally complex mixed-conifer stand. We tested 22 imagery collection methods (altering flight altitude, camera pitch, and image overlap), 12 imagery processing parameterizations, and 286 tree detection methods (algorithms and their parameterizations) to create 7,568 tree maps. We compared these maps to a 3.23-ha ground-truth map of 1,916 trees > 5 m tall that we created using traditional field survey methods.We found that the accuracy of individual tree detection (ITD) and the resulting tree maps was generally maximized by collecting imagery at high altitude (120 m) with at least 90% image-to-image overlap, photogrammetrically processing images into a canopy height model (CHM) with a 2-fold upscaling (coarsening) step, and detecting trees from the CHM using a variable window filter after first applying a moving-window mean smooth to the CHM. Using this combination of methods, we mapped trees with an accuracy that exceeds expectations for our structurally complex forest based on other recent results (for overstory trees > 10 m tall, sensitivity = 0.69 and precision = 0.90). Remotely-measured tree heights corresponded to ground-measured heights with R2 = 0.95. Accuracy was higher for taller trees and lower for understory trees, and it is likely to be higher in lower density and less structurally-complex stands.Our results may guide others wishing to efficiently produce individual-tree maps of conifer forests over broad extents without investing substantial time tailoring imagery acquisition and processing parameters. The resulting tree maps create opportunities for addressing previously intractable ecological questions and increasing the efficiency of forest management.