Combining high-level and low-level visual tasks is a common technique in the field of computer vision. This work integrates the technique of image super resolution to semantic segmentation for document image binarization. It demonstrates that using image super-resolution as a preprocessing step can effectively enhance the results and performance of semantic segmentation.
This paper presents an overview and State-of-the-Art of fully integrated inductors with common fabrication processes used for implementation of these structures into a chip. The first step is an overview of fabrication technologies that starts with standard CMOS general purpose processes expanded by Far-BEOL and substrate alteration process (SOI, Silicon Embedded, TSV, TGV, PTH, Core Insertion); and continues to advanced process nodes like SMMT and Bond Wire utilization. Then, an overview of fully integrated inductor structures consists of selected topologies with notable parameters achieved in last few years. Critical parameters of integrated inductors include: inductance LDC, inductance density LA, quality factor Q, self-resonant frequency FSR and series resistance RDC. These parameters are as important as the purpose and fabrication process.
The paper deals with the problems related to the legal definition of artificial intelligence and at the same time with the issues related to the criminal liability for actions in which artificial intelligence or systems containing it are involved to varying degrees. In this context, artificial intelligence is a key element in the direction of the development of autonomous vehicles. 1
In this article, we discuss handwriting as a behavioral biometric modality. The main contribution focuses on biometric data acquired from handwriting using the pioneering Crayonic KeyVault device, which has a built-in gyroscope, accelerometer, and pressure sensor. Thanks to these sensors, this device can record the movements of the hands while handwriting the digits based on the prompts of the device itself. This data in raw format was sampled and then normalized and parameterized. The time-sampled data were visualized and the intrapersonal similarity was observed from multiple typing attempts of the same participant, and the interpersonal variability when the same number was written from different participants. The data was then divided into a test and a training set. The training set was used to train the machine learning models and the accuracy was verified with the test set. The 91.6% accuracy of the person model identification on the test set was achieved.
Global studies indicate that by 2050, around 70% of the world's population will live, travel and work in urban areas. Cities and suburbs will undergo significant transformations to create the necessary sustainable living conditions for their residents. The main pillars of these transformations - mobility and energy - will require radical adaptation to meet demographic changes and economic growth without increasing congestion and pollution. Cities will need secure, sustainable and affordable mobility and energy solutions that can be connected and integrated with customer-centric infrastructure and services. The creation of next-generation automotive solutions, the implementation of intelligent transport systems, the use of autonomous driving, etc. are a pivotal moment in building the appearance of the new type of cities of the future.[9] In the context of this understanding, in this paper a model of a multi-functional electric scooter with autonomous operation mode, which has been developed and modeled in a 3D environment, is presented. A simulation about the control of the vehicle has been made in Matlab. Regarding the realized design of the scooter, an expert survey was conducted, the results of which are presented in the paper. [1]
