Fish Detection Using Convolutional Neural Networks with Limited Training Data

Abstract. Satellite-Derived Bathymetry (SDB) has been used in many applications related to coastal management. SDB can efficiently fill data gaps obtained from traditional measurements with echo sounding. However, it still requires numerous training data, which is not available in many areas. Furthermore, the accuracy problem still arises considering the linear model could not address the non-relationship between reflectance and depth due to bottom variations and noise. Convolutional Neural Networks (CNN) offers the ability to capture the connection between neighbouring pixels and the non-linear relationship. These CNN characteristics make it compelling to be used for shallow water depth extraction. We investigate the accuracy of different architectures using different window sizes and band combinations. We use Sentinel-2 Level 2A images to provide reflectance values, and Lidar and Multi Beam Echo Sounder (MBES) datasets are used as depth references to train and test the model. A set of Sentinel-2 and in-situ depth subimage pairs are extracted to perform CNN training. The model is compared to the linear transform and applied to two other study areas. Resulting accuracy ranges from 1.3 m to 1.94 m, and the coefficient of determination reaches 0.94. The SDB model generated using a window size of 9x9 indicates compatibility with the reference depths, especially at areas deeper than 15 m. The addition of both short wave infrared bands to the four visible bands in training improves the overall accuracy of SDB. The implementation of the pre-trained model to other study areas provides similar results depending on the water conditions.

Download Full-text

Uncertainty quantification in fault detection using convolutional neural networks

Geophysics ◽

10.1190/geo2020-0424.1 ◽

2021 ◽

pp. 1-45

Author(s):

Runhai Feng ◽

Dario Grana ◽

Niels Balling

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Convolutional Neural Networks ◽

Reservoir Characterization ◽

Epistemic Uncertainty ◽

Confidence Regions ◽

Training Data ◽

Model Parameters ◽

Sigmoid Function ◽

Learning Tools

Segmentation of faults based on seismic images is an important step in reservoir characterization. With the recent developments of deep-learning methods and the availability of massive computing power, automatic interpretation of seismic faults has become possible. The likelihood of occurrence for a fault can be quantified using a sigmoid function. Our goal is to quantify the fault model uncertainty that is generally not captured by deep-learning tools. We propose to use the dropout approach, a regularization technique to prevent overfitting and co-adaptation in hidden units, to approximate the Bayesian inference and estimate the principled uncertainty over functions. Particularly, the variance of the learned model has been decomposed into aleatoric and epistemic parts. The proposed method is applied to a real dataset from the Netherlands F3 block with two different dropout ratios in convolutional neural networks. The aleatoric uncertainty is irreducible since it relates to the stochastic dependency within the input observations. As the number of Monte-Carlo realizations increases, the epistemic uncertainty asymptotically converges and the model standard deviation decreases, because the variability of model parameters is better simulated or explained with a larger sample size. This analysis can quantify the confidence to use fault predictions with less uncertainty. Additionally, the analysis suggests where more training data are needed to reduce the uncertainty in low confidence regions.

Download Full-text

Land Cover Classification from fused DSM and UAV Images Using Convolutional Neural Networks

Remote Sensing ◽

10.3390/rs11121461 ◽

2019 ◽

Vol 11 (12) ◽

pp. 1461 ◽

Cited By ~ 18

Author(s):

Husam A. H. Al-Najjar ◽

Bahareh Kalantar ◽

Biswajeet Pradhan ◽

Vahideh Saeidi ◽

Alfian Abdul Halin ◽

...

Keyword(s):

Neural Networks ◽

Land Cover ◽

Convolutional Neural Networks ◽

Land Cover Classification ◽

Training Data ◽

Surface Model ◽

Kappa Index ◽

Diverse Range ◽

Average Accuracy ◽

Uav Images

In recent years, remote sensing researchers have investigated the use of different modalities (or combinations of modalities) for classification tasks. Such modalities can be extracted via a diverse range of sensors and images. Currently, there are no (or only a few) studies that have been done to increase the land cover classification accuracy via unmanned aerial vehicle (UAV)–digital surface model (DSM) fused datasets. Therefore, this study looks at improving the accuracy of these datasets by exploiting convolutional neural networks (CNNs). In this work, we focus on the fusion of DSM and UAV images for land use/land cover mapping via classification into seven classes: bare land, buildings, dense vegetation/trees, grassland, paved roads, shadows, and water bodies. Specifically, we investigated the effectiveness of the two datasets with the aim of inspecting whether the fused DSM yields remarkable outcomes for land cover classification. The datasets were: (i) only orthomosaic image data (Red, Green and Blue channel data), and (ii) a fusion of the orthomosaic image and DSM data, where the final classification was performed using a CNN. CNN, as a classification method, is promising due to hierarchical learning structure, regulating and weight sharing with respect to training data, generalization, optimization and parameters reduction, automatic feature extraction and robust discrimination ability with high performance. The experimental results show that a CNN trained on the fused dataset obtains better results with Kappa index of ~0.98, an average accuracy of 0.97 and final overall accuracy of 0.98. Comparing accuracies between the CNN with DSM result and the CNN without DSM result for the overall accuracy, average accuracy and Kappa index revealed an improvement of 1.2%, 1.8% and 1.5%, respectively. Accordingly, adding the heights of features such as buildings and trees improved the differentiation between vegetation specifically where plants were dense.

Download Full-text

Beach State Recognition Using Argus Imagery and Convolutional Neural Networks

Remote Sensing ◽

10.3390/rs12233953 ◽

2020 ◽

Vol 12 (23) ◽

pp. 3953

Author(s):

Ashley N. Ellenson ◽

Joshua A. Simmons ◽

Greg W. Wilson ◽

Tyler J. Hesser ◽

Kristen D. Splinter

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

New South ◽

Equilibrium States ◽

The Self ◽

Training Data ◽

Local Data ◽

State Recognition ◽

South Wales ◽

Key Driver

Nearshore morphology is a key driver in wave breaking and the resulting nearshore circulation, recreational safety, and nutrient dispersion. Morphology persists within the nearshore in specific shapes that can be classified into equilibrium states. Equilibrium states convey qualitative information about bathymetry and relevant physical processes. While nearshore bathymetry is a challenge to collect, much information about the underlying bathymetry can be gained from remote sensing of the surfzone. This study presents a new method to automatically classify beach state from Argus daytimexposure imagery using a machine learning technique called convolutional neural networks (CNNs). The CNN processed imagery from two locations: Narrabeen, New South Wales, Australia and Duck, North Carolina, USA. Three different CNN models are examined, one trained at Narrabeen, one at Duck, and one trained at both locations. Each model was tested at the location where it was trained in a self-test, and the single-beach models were tested at the location where it was not trained in a transfer-test. For the self-tests, skill (as measured by the F-score) was comparable to expert agreement (CNN F-values at Duck = 0.80 and Narrabeen = 0.59). For the transfer-tests, the CNN model skill was reduced by 24–48%, suggesting the algorithm requires additional local data to improve transferability performance. Transferability tests showed that comparable F-scores (within 10%) to the self-trained cases can be achieved at both locations when at least 25% of the training data is from each site. This suggests that if applied to additional locations, a CNN model trained at one location may be skillful at new sites with limited new imagery data needed. Finally, a CNN visualization technique (Guided-Grad-CAM) confirmed that the CNN determined classifications using image regions (e.g., incised rip channels, terraces) that were consistent with beach state labelling rules.

Download Full-text

PulseNetOne: Fast Unsupervised Pruning of Convolutional Neural Networks for Remote Sensing

Remote Sensing ◽

10.3390/rs12071092 ◽

2020 ◽

Vol 12 (7) ◽

pp. 1092

Author(s):

David Browne ◽

Michael Giering ◽

Steven Prestwich

Keyword(s):

Remote Sensing ◽

Neural Networks ◽

Deep Learning ◽

Convolutional Neural Networks ◽

Data Augmentation ◽

Recognition Task ◽

Scene Recognition ◽

Training Data ◽

Learning Approach ◽

Scene Classification

Scene classification is an important aspect of image/video understanding and segmentation. However, remote-sensing scene classification is a challenging image recognition task, partly due to the limited training data, which causes deep-learning Convolutional Neural Networks (CNNs) to overfit. Another difficulty is that images often have very different scales and orientation (viewing angle). Yet another is that the resulting networks may be very large, again making them prone to overfitting and unsuitable for deployment on memory- and energy-limited devices. We propose an efficient deep-learning approach to tackle these problems. We use transfer learning to compensate for the lack of data, and data augmentation to tackle varying scale and orientation. To reduce network size, we use a novel unsupervised learning approach based on k-means clustering, applied to all parts of the network: most network reduction methods use computationally expensive supervised learning methods, and apply only to the convolutional or fully connected layers, but not both. In experiments, we set new standards in classification accuracy on four remote-sensing and two scene-recognition image datasets.

Download Full-text

An Efficient Algorithm for Cardiac Arrhythmia Classification Using Ensemble of Depthwise Separable Convolutional Neural Networks

Applied Sciences ◽

10.3390/app10020483 ◽

2020 ◽

Vol 10 (2) ◽

pp. 483 ◽

Cited By ~ 4

Author(s):

Eko Ihsanto ◽

Kalamullah Ramli ◽

Dodi Sudiana ◽

Teddy Surya Gunawan

Keyword(s):

Neural Networks ◽

Feature Extraction ◽

Cardiac Arrhythmia ◽

Convolutional Neural Networks ◽

Computational Cost ◽

Training Data ◽

Qrs Detection ◽

Convolutional Network ◽

Novel Method ◽

Electrocardiogram Ecg

Many algorithms have been developed for automated electrocardiogram (ECG) classification. Due to the non-stationary nature of the ECG signal, it is rather challenging to use traditional handcraft methods, such as time-based analysis of feature extraction and classification, to pave the way for machine learning implementation. This paper proposed a novel method, i.e., the ensemble of depthwise separable convolutional (DSC) neural networks for the classification of cardiac arrhythmia ECG beats. Using our proposed method, the four stages of ECG classification, i.e., QRS detection, preprocessing, feature extraction, and classification, were reduced to two steps only, i.e., QRS detection and classification. No preprocessing method was required while feature extraction was combined with classification. Moreover, to reduce the computational cost while maintaining its accuracy, several techniques were implemented, including All Convolutional Network (ACN), Batch Normalization (BN), and ensemble convolutional neural networks. The performance of the proposed ensemble CNNs were evaluated using the MIT-BIH arrythmia database. In the training phase, around 22% of the 110,057 beats data extracted from 48 records were utilized. Using only these 22% labeled training data, our proposed algorithm was able to classify the remaining 78% of the database into 16 classes. Furthermore, the sensitivity ( S n ), specificity ( S p ), and positive predictivity ( P p ), and accuracy ( A c c ) are 99.03%, 99.94%, 99.03%, and 99.88%, respectively. The proposed algorithm required around 180 μs, which is suitable for real time application. These results showed that our proposed method outperformed other state of the art methods.

Download Full-text

Effects of variability in synthetic training data on convolutional neural networks for 3D head reconstruction

2017 IEEE Symposium Series on Computational Intelligence (SSCI) ◽

10.1109/ssci.2017.8285305 ◽

2017 ◽

Author(s):

Jan Philip Gopfert ◽

Christina Gopfert ◽

Mario Botsch ◽

Barbara Hammer

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

Training Data ◽

Synthetic Training Data

Download Full-text

Convolutional Neural Networks enable efficient, accurate and fine-grained segmentation of plant species and communities from high-resolution UAV imagery

Scientific Reports ◽

10.1038/s41598-019-53797-9 ◽

2019 ◽

Vol 9 (1) ◽

Cited By ~ 20

Author(s):

Teja Kattenborn ◽

Jana Eichel ◽

Fabian Ewald Fassnacht

Keyword(s):

Neural Networks ◽

High Resolution ◽

Convolutional Neural Networks ◽

Low Cost ◽

Training Data ◽

Observation Data ◽

Visual Interpretation ◽

Fine Grained ◽

Wide Range ◽

Vegetation Species

AbstractRecent technological advances in remote sensing sensors and platforms, such as high-resolution satellite imagers or unmanned aerial vehicles (UAV), facilitate the availability of fine-grained earth observation data. Such data reveal vegetation canopies in high spatial detail. Efficient methods are needed to fully harness this unpreceded source of information for vegetation mapping. Deep learning algorithms such as Convolutional Neural Networks (CNN) are currently paving new avenues in the field of image analysis and computer vision. Using multiple datasets, we test a CNN-based segmentation approach (U-net) in combination with training data directly derived from visual interpretation of UAV-based high-resolution RGB imagery for fine-grained mapping of vegetation species and communities. We demonstrate that this approach indeed accurately segments and maps vegetation species and communities (at least 84% accuracy). The fact that we only used RGB imagery suggests that plant identification at very high spatial resolutions is facilitated through spatial patterns rather than spectral information. Accordingly, the presented approach is compatible with low-cost UAV systems that are easy to operate and thus applicable to a wide range of users.

Download Full-text

Dealing with Lack of Training Data for Convolutional Neural Networks: The Case of Digital Pathology

Electronics ◽

10.3390/electronics8030256 ◽

2019 ◽

Vol 8 (3) ◽

pp. 256

Author(s):

Francesco Ponzio ◽

Gianvito Urgese ◽

Elisa Ficarra ◽

Santa Di Cataldo

Keyword(s):

Neural Networks ◽

Transfer Learning ◽

Convolutional Neural Networks ◽

Digital Pathology ◽

Training Data ◽

Automated Classification ◽

Computationally Efficient ◽

Deep Convolutional Neural Networks ◽

Cad Systems ◽

Tissue Characteristics

Thanks to their capability to learn generalizable descriptors directly from images, deep Convolutional Neural Networks (CNNs) seem the ideal solution to most pattern recognition problems. On the other hand, to learn the image representation, CNNs need huge sets of annotated samples that are unfeasible in many every-day scenarios. This is the case, for example, of Computer-Aided Diagnosis (CAD) systems for digital pathology, where additional challenges are posed by the high variability of the cancerous tissue characteristics. In our experiments, state-of-the-art CNNs trained from scratch on histological images were less accurate and less robust to variability than a traditional machine learning framework, highlighting all the issues of fully training deep networks with limited data from real patients. To solve this problem, we designed and compared three transfer learning frameworks, leveraging CNNs pre-trained on non-medical images. This approach obtained very high accuracy, requiring much less computational resource for the training. Our findings demonstrate that transfer learning is a solution to the automated classification of histological samples and solves the problem of designing accurate and computationally-efficient CAD systems with limited training data.

Download Full-text