This paper presents a deep learning-based image texture recognition system. The methodology taken in this solution is formed in a bottom-up manner. It means we swipe a moving window through the image in order to categorize if a given region belongs to one of the classes seen in the training process. This categorization is done based on the Deep Neural Network (DNN) of fixed architecture. The training process is fully automated regarding the training data preparation, investigation of the best training algorithm, and its hyper-parameters. The only human input to the system is the definition of the categories for further recognition and generation of the samples (region markings) in the external application chosen by the user. The system is tested on road surface images where its task is to categorize image regions to a different road category (e.g. curb, road surface damage, etc.) and is featured with 90% and above accuracy.
Three Polish Antarctic Geodynamical Expeditions in 1979/80, 1984/85 and 1987/88 undertook seismic investigations in West Antarctica. Seismic measurements, including multichannel seismic reflection and deep seismic soundings, were carried out in the region of the west coast of the Antarctic Peninsula, between Antarctic Sound and Adelaide Island, Bransfield Strait, South Shetland Islands and South Shetland Trench along several lines with a total length of about 5000 km. Selected crustal sections and one and two-dimensional models of the crust for this area are discussed in detail. The thickness of the crust ranges from 30-33 km in the South Shetland Islands to 38—45 km near the coast of the Antarctic Peninsula. The crustal structure beneath the through of Bransfield Strait is highly anomalous; a seismic discontinuity with velocities of 7.0—7.2 km/s was found at a depth of 10 to 15 km, and a second discontinuity with velocities of about 7.6 km/s was found at a depth of 20—25 km. A seismic inhomogeneity along the Deception-Penguin-Bridgeman volcanic line has also been found. A scheme for the geotectonic division and a geodynamical model of the area are discussed. On the base of all experimental seismic data, it will be possible to construct a continuous geotraverse from Elephant Island, across Bransfield Strait, up to Adelaide Island with a total length of about 1100 km. Crustal section and seismic models along the northern segment of the geotraverse from the King George Island to the Palmer Archipelago are discussed in detail here.
Accurate information on Induction Motor (IM) speed is essential for robust operation of vector controlled IM drives. Simultaneous estimation of speed provides redundancy in motor drives and enables their operation in case of a speed sensor failure. Furthermore, speed estimation can replace its direct measurement for low-cost IM drives or drives operated in difficult environmental conditions. During torque transients when slip frequency is not controlled within the set range of values, the rotor electromagnetic time constant varies due to the rotor deep-bar effect. The model-based schemes for IM speed estimation are inherently more or less sensitive to variability of IM electromagnetic parameters. This paper presents the study on robustness improvement of the Model Reference Adaptive System (MRAS) based speed estimator to variability of IM electromagnetic parameters resulting from the rotor deep-bar effect. The proposed modification of the MRAS-based speed estimator builds on the use of the rotor flux voltage-current model as the adjustable model. The verification of the analyzed configurations of the MRAS-based speed estimator was performed in the slip frequency range corresponding to the IM load adjustment range up to 1.30 of the stator rated current. This was done for a rigorous and reliable assessment of estimators’ robustness to rotor electromagnetic parameter variability resulting from the rotor deep-bar effect. The theoretical reasoning is supported by the results of experimental tests which confirm the improved operation accuracy and reliability of the proposed speed estimator configuration under the considered working conditions in comparison to the classical MRAS-based speed estimator.
Operational amplifies (op amps) are an integral part of many analog and mixed-signal systems. Op amps with vastly different levels of complexity are used to realize functions ranging from DC bias generation to high-speed amplification or filtering. The design of op amps continues to pose a challenge as the supply voltage and transistor channel lengths scale down with each generation of CMOS technologies. The thesis deals with the analysis, design and layout optimization of CMOS op amps in deep Submicron (DSM) from a study case. Finally, layout optimizations of op amps will be given, in which propose optimization techniques to mitigate these DSM effects in the place-and-route stage of VLSI physical design.
In the last few years, a great attention was paid to the deep learning Techniques used for image analysis because of their ability to use machine learning techniques to transform input data into high level presentation. For the sake of accurate diagnosis, the medical field has a steadily growing interest in such technology especially in the diagnosis of melanoma. These deep learning networks work through making coarse segmentation, conventional filters and pooling layers. However, this segmentation of the skin lesions results in image of lower resolution than the original skin image. In this paper, we present deep learning based approaches to solve the problems in skin lesion analysis using a dermoscopic image containing skin tumor. The proposed models are trained and evaluated on standard benchmark datasets from the International Skin Imaging Collaboration (ISIC) 2018 Challenge. The proposed method achieves an accuracy of 96.67% for the validation set .The experimental tests carried out on a clinical dataset show that the classification performance using deep learning-based features performs better than the state-of-the-art techniques.