Background: First case in Indonesia was reported in March 2, 2020 and until November, it has infected more than 400.000 people, with more than 300.000 recovered and 15.000 deaths in total. Especially in Cangkringan District and surrounding area, the latest natural disaster is happening in the midst of a COVID-19 pandemic that is eruption of Mount Merapi which encourage local government and outposts to prepare camps for people nearby who suffered from Mount Merapi eruption. Meanwhile, in evacuation camp, the chance of infected by COVID-19 is high and it is a main problem and a purpose of this research.Methods: Methods that is conducted in this research by obtaining vulnerability level data against the outbreak and making a probability map of virus transmission mainly in the eruption red zone.Result: The outcome of this research is COVID-19 mitigation map within eruption red zone of this regency and recommendations about how to control the outbreak among refugees.Conclusion: Therefore, evacuation can be conducted in the midst of pandemic situation because the transmission case is low.

We developed two machine learning frameworks that could assist in the automated litho-stratigraphic interpretation of seismic volumes without any manual hand labeling from an experienced seismic interpreter. The first framework is an unsupervised hierarchical clustering model to divide seismic images from a volume into certain number of clusters determined by the algorithm. The clustering framework uses a combination of density and hierarchical techniques to determine the size and homogeneity of the clusters. The second framework consists of a self-supervised deep learning framework to label regions of geological interest in seismic images. It projects the latent-space of an encoder-decoder architecture unto two orthogonal subspaces, from which it learns to delineate regions of interest in the seismic images. To demonstrate an application of both frameworks, a seismic volume was clustered into various contiguous clusters, from which four clusters were selected based on distinct seismic patterns: horizons, faults, salt domes and chaotic structures. Images from the selected clusters are used to train the encoder-decoder network. The output of the encoder-decoder network is a probability map of the possibility an amplitude reflection event belongs to an interesting geological structure. The structures are delineated using the probability map. The delineated images are further used to post-train a segmentation model to extend our results to full-vertical sections. The results on vertical sections show that we can factorize a seismic volume into its corresponding structural components. Lastly, we showed that our deep learning framework could be modeled as an attribute extractor and we compared our attribute result with various existing attributes in literature and demonstrate competitive performance with them.

Neuron tracing, as the essential step for neural circuit building and brain information flow analyzing, plays an important role in the understanding of brain organization and function. Though lots of methods have been proposed, automatic and accurate neuron tracing from optical images remains challenging. Current methods often had trouble in tracing the complex tree-like distorted structures and broken parts of neurite from a noisy background. To address these issues, we propose a method for accurate neuron tracing using content-aware adaptive voxel scooping on a convolutional neural network (CNN) predicted probability map. First, a 3D residual CNN was applied as preprocessing to predict the object probability and suppress high noise. Then, instead of tracing on the binary image produced by maximum classification, an adaptive voxel scooping method was presented for successive neurite tracing on the probability map, based on the internal content properties (distance, connectivity, and probability continuity along direction) of the neurite. Last, the neuron tree graph was built using the length first criterion. The proposed method was evaluated on the public BigNeuron datasets and fluorescence micro-optical sectioning tomography (fMOST) datasets and outperformed current state-of-art methods on images with neurites that had broken parts and complex structures. The high accuracy tracing proved the potential of the proposed method for neuron tracing on large-scale.

Abstract With climate change, hydro-climatic hazards, i.e., floods in the Himalayas regions, are expected to worsen, thus, likely to affect humans and socio-economic growth. Precisely, the Koshi River basin (KRB) is often impacted by flooding over the year. However, studies on estimating and predicting floods still lack in this basin. This study aims at developing flood probability map using machine learning algorithms (MLAs): gaussian process regression (GPR) and support vector machine (SVM) with multiple kernel functions including Pearson VII function kernel (PUK), polynomial, normalized poly kernel, and radial basis kernel function (RBF). Historical flood locations with available topography, hydrogeology, and environmental datasets were further considered to build flood model. Two datasets were carefully chosen to measure the feasibility and robustness of MLAs: training dataset (location of floods between 2010 and 2019) and testing dataset (flood locations of 2020) with thirteen flood influencing factors. The validation of the MLAs was evaluated using a validation dataset and statistical indices such as the coefficient of determination (r2: 0.546~0.995), mean absolute error (MAE: 0.009~0.373), root mean square error (RMSE: 0.051~0.466), relative absolute error (RAE: 1.81~88.55%), and root-relative square error (RRSE: 10.19~91.00%). Results showed that the SVM-Pearson VII kernel (PUK) yielded better prediction than other algorithms. The resultant map from SVM-PUK revealed that 27.99% area with low, 39.91% area with medium, 31.00% with high, and 1.10% area with very high probabilities of flooding in the study area. The final flood probability map could add a greatt value to the effort of flood risk mitigation and planning processes in KRB.

Satellite localization often suffers in terms of accuracy due to various reasons. One possible source of errors is represented by the lack of means to eliminate Non-Line-of-Sight satellite-related data. We propose here a method for fusing existing data with new information, extracted by using roof-mounted cameras and adequate image processing algorithms. The roof-mounted camera is used to robustly segment the sky regions. The localization approach can benefit from this new information as it offers a way of excluding the Non-Line-of-Sight satellites. The output of the camera module is a probability map. One can easily decide which satellites should not be used for localization, by manipulating this probability map. Our approach is validated by extensive tests, which demonstrate the improvement of the localization itself (Horizontal Positioning Error reduction) and a moderate degradation of Horizontal Protection Level due to the Dilution of Precision phenomenon, which appears as a consequence of the reduction of the satellites’ number used for localization.

This paper proposes a better semi-supervised semantic segmentation network using an improved generative adversarial network. It is important for the discriminator on the pixel level to know whether it correctly distinguishes the predicted probability map. However, currently there is no correlation between the actual credibility and the confidence map generated by the pixel-level discriminator. We study this problem and a new network is proposed, which includes one generator and two discriminators. One of the discriminators can output more reliable confidence maps on the pixel level and the other is trained to generate the probability on the image level, which is used as the dynamic threshold in the semi-supervised module instead of being set manually. In addition, the trusted region shared by the two discriminators is used to provide the semi-supervised reference. Through experiments on the PASCAL VOC 2012 and Cityscapes datasets, the proposed network brings better gains, proving the effectiveness of the network.