Computer Vision and Deep Learning Techniques for the Analysis of Drone-Acquired Forest Images, a Transfer Learning Study

Kentsch, Sarah; Caceres, Maximo Larry Lopez; Serrano, Daniel; Roure, Ferran; Diéz, Yago

doi:10.3390/rs12081287

Cited by 65 publications

(42 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this study, DNNs were used to locate and identify the six classes that are defined in Section 2.3 , with a focus set on the blueberry class. The basis for this deep learning approach is described in previous studies [ 20 , 43 ], which led to the use of the algorithms that are described in this section.…”

Section: Methodsmentioning

confidence: 99%

“…is needed to capture all of the possible image variabilities. However, as proven by our previous work [ 20 , 43 ], transfer learning is a useful tool for image analysis applications, where the training dataset is too small to properly train these feature extractors from scratch.…”

Section: Methodsmentioning

confidence: 99%

“…Wetlands are protected environments with limited ground accessibility making UAVs particularly appropriate for data collection. UAVs offer the possibility to cover large areas with high resolution images, and they have proved their usefulness in a variety of studies in agriculture [ 16 , 17 ] and forestry [ 18 , 19 , 20 ]. Still, UAV images present challenges like the pre-pocessing of the data.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Analysis of UAV-Acquired Wetland Orthomosaics Using GIS, Computer Vision, Computational Topology and Deep Learning

Kentsch

Cabezas

Tomhave

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

Invasive blueberry species endanger the sensitive environment of wetlands and protection laws call for management measures. Therefore, methods are needed to identify blueberry bushes, locate them, and characterise their distribution and properties with a minimum of disturbance. UAVs (Unmanned Aerial Vehicles) and image analysis have become important tools for classification and detection approaches. In this study, techniques, such as GIS (Geographical Information Systems) and deep learning, were combined in order to detect invasive blueberry species in wetland environments. Images that were collected by UAV were used to produce orthomosaics, which were analysed to produce maps of blueberry location, distribution, and spread in each study site, as well as bush height and area information. Deep learning networks were used with transfer learning and unfrozen weights in order to automatically detect blueberry bushes reaching True Positive Values (TPV) of 93.83% and an Overall Accuracy (OA) of 98.83%. A refinement of the result masks reached a Dice of 0.624. This study provides an efficient and effective methodology to study wetlands while using different techniques.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Analysis of UAV-Acquired Wetland Orthomosaics Using GIS, Computer Vision, Computational Topology and Deep Learning

Kentsch

Cabezas

Tomhave

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…The experimental results showed that the highest accuracy of trunks and leaves based on transfer learning was improved by 51.38% and 51.69%, respectively, compared with the accuracy of ordinary deep learning. In 2020, Kentsch et al conducted research on classification identification of a winter orthomosaic dataset by ResNet-50 based on transfer learning [16]. The results showed that from no transfer learning to transfer learning from a general-purpose dataset, the accuracy with a 9.78% improvement.…”

Section: Introductionmentioning

confidence: 99%

A Transfer Residual Neural Network Based on ResNet-34 for Detection of Wood Knot Defects

Gao

et al. 2021

Forests

View full text Add to dashboard Cite

In recent years, due to the shortage of timber resources, it has become necessary to reduce the excessive consumption of forest resources. Non-destructive testing technology can quickly find wood defects and effectively improve wood utilization. Deep learning has achieved significant results as one of the most commonly used methods in the detection of wood knots. However, compared with convolutional neural networks in other fields, the depth of deep learning models for the detection of wood knots is still very shallow. This is because the number of samples marked in the wood detection is too small, which limits the accuracy of the final prediction of the results. In this paper, ResNet-34 is combined with transfer learning, and a new TL-ResNet34 deep learning model with 35 convolution depths is proposed to detect wood knot defects. Among them, ResNet-34 is used as a feature extractor for wood knot defects. At the same time, a new method TL-ResNet34 is proposed, which combines ResNet-34 with transfer learning. After that, the wood knot defect dataset was applied to TL-ResNet34 for testing. The results show that the detection accuracy of the dataset trained by TL-ResNet34 is significantly higher than that of other methods. This shows that the final prediction accuracy of the detection of wood knot defects can be improved by TL-ResNet34.

show abstract

“…Traditional image is an important carrier of information in multimedia, and 70% of the information people obtained by comes from the visual system [4], [5]. However, an effective processing method is urgently needed because of the large amount of digital image data and considerable storage space [6]- [8].…”

Section: Introductionmentioning

confidence: 99%

Demonstration of Quantum Image Edge Extration Enhancement Through Improved Sobel Operator

Chu

2020

IEEE Access

View full text Add to dashboard Cite

Quantum image processing is a system where both quantum computing and image processings are key. Now it is required to investigate how to apply image processing concepts such as image edge detection and improved Sobel operator, to quantum image processing. Just as traditional image processing helps to image manipulation, our efforts in quantum image processing contribute to the further development of quantum algorithm and quantum information theory. Edge detection is an important problem in traditional image processing, and image edge detection algorithm can filter image edge features and retain important attributes. We proposed a quantum edge detection algorithm, which employs improved Sobel operator to improve the performance, so as to solve the problem of unsatisfactory traditional image edge detection methods. In practical application, the amount of image data to be processed increases sharply, and the computing power of classical computer becomes a limitation. Quantum information processing can effectively accelerate many classical problems by virtue of quantum mechanical characteristics, such as quantum superposition, entanglement, parallelism. Our proposed method uses the novel enhanced quantum representation to store quantum images, which stores all the pixels in the image in a superimposed state, realizing parallel computation. The improved eight-direction Sobel operator is used to calculate the gray gradient, and the quantum circuit is designed to realize quantum edge detection. Simulation results have shown that our algorithm can extract edge with dimension 2 n × 2 n when the computational complexity is O(n 2 + 2 q+3). The improved algorithm can detect more edge details and has strong adaptability.

show abstract

Computer Vision and Deep Learning Techniques for the Analysis of Drone-Acquired Forest Images, a Transfer Learning Study

Cited by 65 publications

References 22 publications

Analysis of UAV-Acquired Wetland Orthomosaics Using GIS, Computer Vision, Computational Topology and Deep Learning

Analysis of UAV-Acquired Wetland Orthomosaics Using GIS, Computer Vision, Computational Topology and Deep Learning

A Transfer Residual Neural Network Based on ResNet-34 for Detection of Wood Knot Defects

Demonstration of Quantum Image Edge Extration Enhancement Through Improved Sobel Operator

Contact Info

Product

Resources

About