Deep neural networks are currently the focus of many remote sensing approaches related to forest management. Although they return satisfactory results in most tasks, some challenges related to ...hyperspectral data remain, like the curse of data dimensionality. In forested areas, another common problem is the highly-dense distribution of trees. In this paper, we propose a novel deep learning approach for hyperspectral imagery to identify single-tree species in highly-dense areas. We evaluated images with 25 spectral bands ranging from 506 to 820 nm taken over a semideciduous forest of the Brazilian Atlantic biome. We included in our network’s architecture a band combination selection phase. This phase learns from multiple combinations between bands which contributed the most for the tree identification task. This is followed by a feature map extraction and a multi-stage model refinement of the confidence map to produce accurate results of a highly-dense target. Our method returned an f-measure, precision and recall values of 0.959, 0.973, and 0.945, respectively. The results were superior when compared with a principal component analysis (PCA) approach. Compared to other learning methods, ours estimate a combination of hyperspectral bands that most contribute to the mentioned task within the network’s architecture. With this, the proposed method achieved state-of-the-art performance for detecting and geolocating individual tree-species in UAV-based hyperspectral images in a complex forest.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
In recent years, many agriculture-related problems have been evaluated with the integration of artificial intelligence techniques and remote sensing systems. Specifically, in fruit detection ...problems, several recent works were developed using Deep Learning (DL) methods applied in images acquired in different acquisition levels. However, the increasing use of anti-hail plastic net cover in commercial orchards highlights the importance of terrestrial remote sensing systems. Apples are one of the most highly-challenging fruits to be detected in images, mainly because of the target occlusion problem occurrence. Additionally, the introduction of high-density apple tree orchards makes the identification of single fruits a real challenge. To support farmers to detect apple fruits efficiently, this paper presents an approach based on the Adaptive Training Sample Selection (ATSS) deep learning method applied to close-range and low-cost terrestrial RGB images. The correct identification supports apple production forecasting and gives local producers a better idea of forthcoming management practices. The main advantage of the ATSS method is that only the center point of the objects is labeled, which is much more practicable and realistic than bounding-box annotations in heavily dense fruit orchards. Additionally, we evaluated other object detection methods such as RetinaNet, Libra Regions with Convolutional Neural Network (R-CNN), Cascade R-CNN, Faster R-CNN, Feature Selective Anchor-Free (FSAF), and High-Resolution Network (HRNet). The study area is a highly-dense apple orchard consisting of Fuji Suprema apple fruits (Malus domestica Borkh) located in a smallholder farm in the state of Santa Catarina (southern Brazil). A total of 398 terrestrial images were taken nearly perpendicularly in front of the trees by a professional camera, assuring both a good vertical coverage of the apple trees in terms of heights and overlapping between picture frames. After, the high-resolution RGB images were divided into several patches for helping the detection of small and/or occluded apples. A total of 3119, 840, and 2010 patches were used for training, validation, and testing, respectively. Moreover, the proposed method’s generalization capability was assessed by applying simulated image corruptions to the test set images with different severity levels, including noise, blurs, weather, and digital processing. Experiments were also conducted by varying the bounding box size (80, 100, 120, 140, 160, and 180 pixels) in the image original for the proposed approach. Our results showed that the ATSS-based method slightly outperformed all other deep learning methods, between 2.4% and 0.3%. Also, we verified that the best result was obtained with a bounding box size of 160 × 160 pixels. The proposed method was robust regarding most of the corruption, except for snow, frost, and fog weather conditions. Finally, a benchmark of the reported dataset is also generated and publicly available.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
•In this paper, a new method is proposed for estimating soybean defoliation.•Regression and convolutional neural networks were employed in our method.•We also presented three approaches to generate ...synthetic images.•The proposed method was trained only with synthetic images.•Experiments on real images showed excellent results.
Agriculture plays an important role in the economy of several countries, contributing from the production of food and income to the generation of jobs. To improve productivity in agriculture, proper crop management should be accomplished through pest control. One approach is to monitor the defoliation level, that is, the percentage of leaf damaged by insects. Despite the importance, this monitoring is performed manually most of the time, which affects reliability as well as it is considered to be a time-consuming task. In this paper, we propose a new fully-automatic method to estimate the defoliation level based on convolutional neural networks (CNN). The main CNN architectures (AlexNet, VGGNet and ResNet) were adapted from classification to regression by replacing the softmax layer with a fully-connected layer with one neuron and a sigmoid activation function. Since CNNs require a large number of training examples, this paper also proposes approaches for generating synthetic defoliation images. In this way, our method is trained only with synthetic images and evaluated using real images. In the experiments, we obtained a root mean square error of only 4.57 even for images with severe defoliation. Additionally, we presented experimental evidence that the proposed method reconstructs the damaged leaf parts to then estimate the defoliation level.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
Accurately mapping croplands is an important prerequisite for precision farming since it assists in field management, yield-prediction, and environmental management. Crops are sensitive to planting ...patterns and some have a limited capacity to compensate for gaps within a row. Optical imaging with sensors mounted on Unmanned Aerial Vehicles (UAV) is a cost-effective option for capturing images covering croplands nowadays. However, visual inspection of such images can be a challenging and biased task, specifically for detecting plants and rows on a one-step basis. Thus, developing an architecture capable of simultaneously extracting plant individually and plantation-rows from UAV-images is yet an important demand to support the management of agricultural systems. In this paper, we propose a novel deep learning method based on a Convolutional Neural Network (CNN) that simultaneously detects and geolocates plantation-rows while counting its plants considering highly-dense plantation configurations. The experimental setup was evaluated in (a) a cornfield (Zea mays L.) with different growth stages (i.e. recently planted and mature plants) and in a (b) Citrus orchard (Citrus Sinensis Pera). Both datasets characterize different plant density scenarios, in different locations, with different types of crops, and from different sensors and dates. This scheme was used to prove the robustness of the proposed approach, allowing a broader discussion of the method. A two-branch architecture was implemented in our CNN method, where the information obtained within the plantation-row is updated into the plant detection branch and retro-feed to the row branch; which are then refined by a Multi-Stage Refinement method. In the corn plantation datasets (with both growth phases – young and mature), our approach returned a mean absolute error (MAE) of 6.224 plants per image patch, a mean relative error (MRE) of 0.1038, precision and recall values of 0.856, and 0.905, respectively, and an F-measure equal to 0.876. These results were superior to the results from other deep networks (HRNet, Faster R-CNN, and RetinaNet) evaluated with the same task and dataset. For the plantation-row detection, our approach returned precision, recall, and F-measure scores of 0.913, 0.941, and 0.925, respectively. To test the robustness of our model with a different type of agriculture, we performed the same task in the citrus orchard dataset. It returned an MAE equal to 1.409 citrus-trees per patch, MRE of 0.0615, precision of 0.922, recall of 0.911, and F-measure of 0.965. For the citrus plantation-row detection, our approach resulted in precision, recall, and F-measure scores equal to 0.965, 0.970, and 0.964, respectively. The proposed method achieved state-of-the-art performance for counting and geolocating plants and plant-rows in UAV images from different types of crops. The method proposed here may be applied to future decision-making models and could contribute to the sustainable management of agricultural systems.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZRSKP
Pantanal is the largest continuous wetland in the world, but its biodiversity is currently endangered by catastrophic wildfires that occurred in the last three years. The information available for ...the area only refers to the location and the extent of the burned areas based on medium and low-spatial resolution imagery, ranging from 30 m up to 1 km. However, to improve measurements and assist in environmental actions, robust methods are required to provide a detailed mapping on a higher-spatial scale of the burned areas, such as PlanetScope imagery with 3–5 m spatial resolution. As state-of-the-art, Deep Learning (DL) segmentation methods, in specific Transformed-based networks, are one of the best emerging approaches to extract information from remote sensing imagery. Here we combine Transformers DL methods and high-resolution planet imagery to map burned areas in the Brazilian Pantanal wetland. We first compared the performances of multiple DL-based networks, namely Segformer and DTP Transformers methods with CNN-based networks like PSPNet, FCN, DeepLabV3+, OCRNet, and ISANet, applied in Planet imagery, considering RGB and near-infrared within a large dataset of 1282 image patches (512 × 512 pixels). We later verified the generalization capability of the model for segmenting burned areas in different areas, located in the Brazilian Amazon, which is also worldwide known due to its environmental relevance. As a result, the two transformers based-methods, SegFormer (F1-score equals 95.91%) and DTP (F1-score equals 95.15%), provided the most accurate results in mapping burned forest areas in Pantanal. Results show that the combination of SegFormer and RGB+NIR image with pre-trained weights is the best option (F1-score of 96.52%) to distinguish burned from not-burned areas. When applying the generated model in two Brazilian Amazon forest regions, we achieved F1-score averages of 95.88% for burned areas. We conclude that Transformer-based networks are fit to deal with burned areas in two of the most relevant environmental areas of the world using high-spatial-resolution imagery.
•Pre-trained weights impact the overall accuracy to map burned and not burned areas.•SegFormer combined with RGB+NIR imagery is the best model to segment burned areas.•Transformer DL models present high generalization capability in mapping burned areas.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Tree species mapping is an important type of information demanded in different study fields. However, this task can be expensive and time-consuming, making it difficult to monitor extensive areas. ...Hence, automatic methods are required to optimize tree species mapping. Here, we propose a deep learning-based mobile application tool for tree species classification in high-spatial-resolution RGB images. Several deep learning architectures were evaluated, including mobile networks and traditional models. A total of 2,349 images were used, of which 1,174 images consisted of the Dipteryx alata species and 1,175 images of other local species. These images were manually annotated and randomly divided into training (70%), validation (20%), and testing (10%) subsets, considering the five-fold cross-validation. We evaluated the accuracy and speed (GPU and CPU) of all the implemented deep learning architectures. We found out that the traditional networks have the best performance in terms of F1 score; however, mobile networks are faster. Inception V3 model achieved the best accuracy (F1 score of 97.4%), and MobileNet the worst (F1 score of 83.84%). The MobileNet obtained the best classification speed for CPU (with a mean execution time of 102.8 ms) and GPU (72.4 ms) units. For comparison, Inception V3 achieved a mean execution time of 1058.3 ms for CPU and 634.5 ms for GPU. We conclude that the mobile application proposed can be successfully used to run mobile networks and traditional networks for image classification, but the balance between accuracy and execution time needs to be carefully assessed. This mobile app is a tool for researchers, policymakers, non-governmental organizations, and the general public who intends to assess the tree species, providing a GUI-based platform for non-programmers to access the capabilities of deep learning models in complex classification tasks.
•Deep learning models executed in mobile applications can classify tree species in RGB.•Traditional networks are better (F1-score of 97.4%) than mobile (83.8%) networks.•Mobile networks are faster than traditional models in the mean execution time on task.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
This paper presents a Convolutional Neural Network (CNN) approach for counting and locating objects in high-density imagery. To the best of our knowledge, this is the first object counting and ...locating method based on a feature map enhancement combined with a multi-sigma refinement of the confidence map. The proposed method was evaluated in two counting datasets: trees and cars. For the tree dataset, our method returned a mean absolute error (MAE) of 2.05, a root-mean-squared error (RMSE) of 2.87 and a coefficient of determination (R2) of 0.986. For the car dataset (CARPK and PUCPR+), our method was superior to state-of-the-art methods. In the these datasets, our approach achieved an MAE of 4.45 and 3.16, an RMSE of 6.18 and 4.39, and an R2 of 0.975 and 0.999, respectively. We conclude that the proposed method is suitable for dealing with high object-density, returning a state-of-the-art performance for counting and locating objects.
•We propose a new method for counting and locating objects with high-density.•Our method proposes a multi-stage refinement to improve location in dense regions.•Results using the proposed method proved superior to traditional approaches.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Fingerling counting is an important task for decision-making in the aquaculture context. The counting is usually performed by a human, which is time-consuming and prone to errors. Artificial ...intelligence methods applied to image interpretation can be a great strategy for solving this task automatically. However, applying machine learning to attend to aquaculture issues is an underexplored field that requires novel investigations, especially of methods that explore temporal information in videos. In this study, we propose a new method to locate and count fingerlings in a sequence of images using convolutional neural networks. The proposed method estimates three tasks in a multi-task approach. The first task consists of predicting the probability of a fingerling occurring in each pixel of the frame, while the second and third tasks estimate the movement performed by the fingerlings. Motion prediction is used as a complement to fingerling detection, including relevant information especially when two or more fingerlings are in contact. Experimental results indicated that the use of temporal information considerably increases the results, reaching F1 of 97.89. The proposed method was evaluated in frames with different numbers of fingerlings (from 0 to 10) and all obtained relevant results, with an F1 of 95.42 or higher. The study also showed that, in most cases, the proposed method can detect the contact of two or more fingerlings, which is considered the main challenge of the detection and counting of fingerlings.
•CNNs show relevant results to locate and count fingerlings in sequence of images.•The use of temporal information considerably increases the results.•The method can detect the joining of two fingerlings visually forming a single one.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
•We propose a loss function for semantic segmentation methods in vegetation.•We address class imbalance and labeling uncertainty.•Results using the proposed method proved superior to traditional ...approaches.
Recently, Convolutional Neural Networks (CNN) methods achieved impressive success in semantic segmentation tasks. However, challenges like class imbalance around samples and the uncertainty in human pixel-labeling are not completely addressed. Here we present an approach that calculates a weight for each pixel considering its class and uncertainty during the labeling process. The pixel-wise weights are used at the training phase to increase or decrease the importance of the pixels accordingly. Experimental results were conducted adapting well-known CNN methods FCN and SegNet; however, this strategy can be applied to any segmentation method. We evaluated the experiments for semantic segmentation of urban trees in aerial imageries. The robustness of the approach was assessed using a dataset with terrestrial images from vegetation with a drastic imbalance condition. We achieved significant improvements in the tasks compared to the baseline methods. We also verified that the proposed strategy proved to be more invariant to noise. The approach presented in this paper could be used within a wide range of semantic segmentation methods to improve their robustness.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Local descriptors for soybean disease recognition Pires, Rillian Diello Lucas; Gonçalves, Diogo Nunes; Oruê, Jonatan Patrick Margarido ...
Computers and electronics in agriculture,
July 2016, 2016-07-00, 20160701, Volume:
125
Journal Article
Peer reviewed
•A novel approach is proposed for soybean disease recognition using leaf images.•It is based on local descriptors and bag-of-visual words.•Experimental results on an image dataset with 1200 samples ...validate its effectiveness.•Results also show that color increases the correct classification rate.
The detection of diseases is of vital importance to increase the productivity of soybean crops. The presence of the diseases is usually conducted visually, which is time-consuming and imprecise. To overcome these issues, there is a growing demand for technologies that aim at early and automated disease detection. In this line of work, we introduce an effective (over 98% of accuracy) and efficient (an average time of 0.1s per image) method to computationally detect soybean diseases. Our method is based on image local descriptors and on the summarization technique Bag of Visual Words. We tested our approach on a dataset composed of 1200 scanned soybean leaves considering healthy samples, and samples with evidence of three diseases commonly observed in soybean crops – Mildew, Rust Tan, and Rust RB. The experimental results demonstrated the accuracy of the proposed approach and suggested that it can be easily applied to other kinds of crops.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP