In geospatial applications such as urban planning and land use management, automatic detection and classification of earth objects are essential and primary subjects. When the significant semantic segmentation algorithms are considered, DeepLabV3+ stands out as a state-of-the-art CNN. Although the DeepLabV3+ model is capable of extracting multi-scale contextual information, there is still a need for multi-stream architectural approaches and different training approaches of the model that can leverage multi-modal geographic datasets. In this study, a new end-to-end dual-stream architecture that considers geospatial imagery was developed based on the DeepLabV3+ architecture. As a result, the spectral datasets other than RGB provided increments in semantic segmentation accuracies when they were used as additional channels to height information. Furthermore, both the given data augmentation and Tversky loss function which is sensitive to imbalanced data accomplished better overall accuracies. Also, it has been shown that the new dual-stream architecture using Potsdam and Vaihingen datasets produced 88.87% and 87.39% overall semantic segmentation accuracies, respectively. Eventually, it was seen that enhancement of the traditional significant semantic segmentation networks has a great potential to provide higher model performances, whereas the contribution of geospatial data as the second stream to RGB to segmentation was explicitly shown.

Subtitle data carried by MPEG-2 transport stream includes texts of dialogs in bitmap graphics format. However, this data is not reachable for the blind audience. In addition, carrying extra compatible data with subtitle packets can be helpful for testing subtitle software automatically and for presenting live content (i.e. latest news or special advertisements). In this paper, we propose a method for converting subtitle data into text format by recognizing characters using neural networks. The converted data can be used for blind television users and testing subtitle software automatically. We also define a new subtitle segment for carrying test comparison data and a new subtitle page for representing live content. We present our implementation on Linux platform.

Relational database management systems have been used for storing data for a long time. However, these systems are insufficient to analyze the large and complex structure of the data. Graph databases are becoming more common day by day due to their capacity to contribute to the analysis. Also, graph databases are better at modeling and querying complex relationships than relational databases. To use graph databases with old data stored in relational databases a transfer process is needed. In this study, the problems to be encountered in transferring the data stored in a relational database to a graph database were examined and methods that could be used as solutions to them were proposed. In addition, it is aimed to prevent data loss and data inconsistency that may occur with design errors in relational databases. For this purpose, the normalization process needs to be applied to a relational database before transferring data to a graph database. In our study, we developed a method that converts data to the first normal form during the transfer. But for better data consistency in practice third normal form is the minimum requirement. By using the functional dependencies found, it is possible to make relational databases suitable for higher normal forms. For functional dependency detection, which is normally a very time-consuming and costly process, we developed a method based on a graph database.

Boundary extraction in remote sensing has an important task in studies such as environmental observa-tion, risk management and monitoring urban growth. Although significant progress has been made in the different calculation methods proposed, there are issues that need improvement, especially in terms of accuracy, efficiency and speed. In this study, dual stream network architecture of three different models that can obtain boundary extraction by using normalized Digital Surface Model (nDSM), Normalized Difference Vegetation Index (NDVI) and Near-Infrared (IR) band as the second stream, was explained. Model I is designed as the original HED, whereas the second stream of Model II, III, and IV use nDSM, nDSM + NDVI and nDSM + NDVI + IR, respectively. Thus, by comparing the models trained based on different data combinations, the contribution of different input data to the success of boundary extraction was revealed. For the training of the models, boundary maps produced from The International Society for Photogrammetry and Remote Sensing (ISPRS) Potsdam data set and input datasets augmented by rotation, mirroring and rotation were used. When the test results obtained from two-stream and multi-data-based models are evaluated, 11% better accuracy has achieved with Model IV compared to the original HED. The outcomes clearly revealed the importance of using multispectral band, height data and vegetation information as input data in boundary extraction beside commonly used RGB images.

Continuous monitoring of livestock operations is vitally important for a sustainable production system. Monitoring systems based on cameras are not sufficient in livestock barns since they require visual inspection and ignore vocal conditions within the barn. These systems are also quite expensive for most small family operations. A prototype device that costed $ 470 was developed to remotely monitor the barn based on sound sensors (microphones) data. This device also warns the operator by sending an SMS at sound intensities exceeding the predetermined durations and threshold values. It also makes it possible to listen to the barn by phone if needed. The device and associated web database was tested in this study. The main challenge was the determination of threshold values at which sensors are to generate warning SMS messages. As a method, Z-score of 2.33 which corresponds to area left of the 99% of the normally distributed data curve was determined representing the highest values with a possibility of 1% observation for each sensor. An average value of 97 dB was determined to be a threshold suggestion for future studies. A customizable web-based MySQL database was created to monitor and evaluate the long term data collected via the system.