No description available.

With the rapid worldwide spread of Coronavirus (COVID-19 and COVID-20), wearing face masks in public becomes a necessity to mitigate the transmission of this or other pandemics. However, with the lack of on-ground automated prevention measures, depending on humans to enforce face mask-wearing policies in universities and other organizational buildings, is a very costly and time-consuming measure. Without addressing this challenge, mitigating highly airborne transmittable diseases will be impractical, and the time to react will continue to increase. Considering the high personnel traffic in buildings and the effectiveness of countermeasures, that is, detecting and offering unmasked personnel with surgical masks, our aim in this paper is to develop automated detection of unmasked personnel in public spaces in order to respond by providing a surgical mask to them to promptly remedy the situation. Our approach consists of three key components. The first component utilizes a deep learning architecture that integrates deep residual learning (ResNet-50) with Feature Pyramid Network (FPN) to detect the existence of human subjects in the videos (or video feed). The second component utilizes Multi-Task Convolutional Neural Networks (MT-CNN) to detect and extract human faces from these videos. For the third component, we construct and train a convolutional neural network classifier to detect masked and unmasked human subjects. Our techniques were implemented in a mobile robot, Thor, and evaluated using a dataset of videos collected by the robot from public spaces of an educational institute in the U.S. Our evaluation results show that Thor is very accurate achieving an F_{1} score of 87.7% with a recall of 99.2% in a variety of situations, a reasonable accuracy given the challenging dataset and the problem domain.

<p>Istraživanja predstavljena u disertaciji imala su za cilj razvoj nove metode bazirane na dubokim konvolucijskim neuoronskim mrežama u cilju detekcije bolesti biljaka na osnovu slike lista. U okviru eksperimentalnog dela rada prikazani su dosadašnji literaturno dostupni pristupi u automatskoj detekciji bolesti biljaka kao i ograničenja ovako dobijenih modela kada se koriste u prirodnim uslovima. U okviru disertacije uvedena je nova baza slika listova, trenutno najveća po broju slika u poređenju sa javno dostupnim bazama, potvrđeni su novi pristupi augmentacije bazirani na GAN arhitekturi nad slikama listova uz novi specijalizovani dvo-koračni pristup kao potencijalni odgovor na nedostatke postojećih rešenja.</p> / <p>The research presented in this thesis was aimed at developing a novel method based on deep convolutional neural networks for automated plant disease detection. Based on current available literature, specialized two-phased deep neural network method introduced in the experimental part of thesis solves the limitations of state-of-the-art plant disease detection methods and provides the possibility for a practical usage of the newly developed model. In addition, a new dataset was introduced, that has more images of leaves than other publicly available datasets, also GAN based augmentation approach on leaves images is experimentally confirmed.</p>

A big hurdle for many companies to start using machine learning is that trending techniques need a huge amount of structured data. One potential way to reduce the need for data is taking advantage of previous knowledge from a related task. This is so called transfer learning. A basic description of it would be when you take a model trained on existing data and reuse that for another problem. The purpose of this master thesis is to investigate if transfer learning can reduce the need for data when faced with a new machine learning task which is, in particular, to use transfer learning on ultrasound spectrograms of weld joints for predictive maintenance. The base for transfer learning is VGGish, a convolutional neural network model trained on audio samples collected from YouTube videos. The pre-trained weights are kept, and the prediction layer is replaced with a new prediction layer consisting of two neurons. The whole model is re-trained on the ultrasound spectrograms. The dataset is restricted to a minimum of ten and a maximum of 100 training samples. The results are evaluated and compared to a regular convolutional neural network trained on the same data. The results show that transfer learning improves the test accuracy compared to the regular convolutional neural network when the dataset is small. This thesis project concludes that transfer learning can reduce the need for data when faced with a new machine learning task. The results indicate that transfer learning could be useful in the industry.

Neutrinoless Double Beta Decay(0𝜈𝛽𝛽) is one of the major research interests in neutrino physics. The discovery of 0𝜈𝛽𝛽 would answer persistent puzzles in the Standard Model of Elementary Particles. KamLAND-Zen is one of the leading efforts in the search of 0𝛽𝛽 and has acquired data from 745 kg of ^{136}Xe over 224 live-days. This data is analyzed using a Bayesian approach consisting of a Markov Chain Monte Carlo (MCMC) algorithm. The implementation of the Bayesian analysis, which is the focal point of this dissertation, yields a 90\% Credible Interval at T^{0𝜈}_{1/2} = 7.03 × 10^{25} years. Finally, a machine learning event classification algorithm, based on a spherical convolutional neural network (spherical CNN) was developed to increase the T^{0𝜈}_{1/2} sensitivity. The classification power of this algorithm was demonstrated on a Monte Carlo detector simulation, and a data driven classifier was trained to reject crucial backgrounds in the 0𝜈𝛽𝛽 analysis. After implementing the spherical CNN, an increase in T^{0𝜈}_{1/2} sensitivity of 11.0% is predicted. These early studies pave the way for substantial improvements in future 0𝜈𝛽𝛽 analyses.

<div> <div> <div> <p>In this thesis, two aspects of video coding improvement are discussed, namely error resilience and coding efficiency. </p> <p>With the increasing amount of videos being created and consumed, better video compression tools are needed to provide reliable and fast transmission. Many popular video coding standards such as VPx, H.26x achieve video compression by using spa- tial and temporal dependencies in the source video signal. This makes the encoded bitstream vulnerable to errors during transmission. In this thesis, we investigate an error resilient video coding for the VP9 bitstreams using error resilience packets. An error resilient packet consists of encoded keyframe contents and the prediction sig- nals for each non-keyframe. Experimental results exhibit that our proposed method is effective under typical packet loss conditions. </p> <p>In the second part of the thesis, we first present an automatic stillness feature detection method for group of pictures. The encoder adaptively chooses the coding structure for each group of pictures based on its stillness feature to optimize the coding efficiency. </p> <p>Secondly, a content-based video coding method is proposed. Modern video codecs including the newly developed AOM/AV1 utilize hybrid coding techniques to remove spatial and temporal redundancy. However, the efficient exploitation of statistical dependencies measured by a mean squared error (MSE) does not always produce the best psychovisual result. One interesting approach is to only encode visually relevant information and use a different coding method for “perceptually insignificant” regions </p> </div> </div> <div> <div> <p>xiv </p> </div> </div> </div> <div> <div> <div> <p>in the frame. In this thesis, we introduce a texture analyzer before encoding the input sequences to identify detail irrelevant texture regions in the frame using convolutional neural networks. The texture region is then reconstructed based on one set of motion parameters. We show that for many standard test sets, the proposed method achieved significant data rate reductions. </p> </div> </div> </div>

Nowadays, everyone is very busy and running around trying to maintain a balance between their work life and family, as the working hours are increasing day by day. In such hassled life people either ignore or do not give enough attention to a healthy diet. An imperative part of a healthy eating routine is the cognizance and maintenance of nourishing data and comprehension of how extraordinary sustenance and nutritious constituents influence our bodies. Besides in the USA, in many other countries, nutritional information is fundamentally passed on to consumers through nutrition labels (NLs) which can be found in all packaged food products in the form of nutrition table. However, sometimes it turns out to be challenging to utilize this information available in these NLs notwithstanding for consumers who are health conscious as they may not be familiar with nutritional terms and discover it hard to relate nutritional information into their day by day activities because of lack of time, inspiration, or training. So it is essential to automate this information gathering and interpretation procedure by incorporating Machine Learning based algorithm to abstract nutritional information from NLs on the grounds that it enhances the consumer’s capacity to participate in nonstop nutritional information gathering and analysis.

Colony Collapse Disorder (CCD) has been a major threat to bee colonies around the world which affects vital human food crop pollination. The decline in bee population can have tragic consequences, for humans as well as the bees and the ecosystem. Bee health has been a cause of urgent concern for farmers and scientists around the world for at least a decade but a specific cause for the phenomenon has yet to be conclusively identified. This work uses Artificial Intelligence and Computer Vision approaches to develop and analyze techniques to help in continuous monitoring of bee traffic which will further help in monitoring forager traffic. Bee traffic is the number of bees moving in a given area in front of the hive over a given period of time. And, forager traffic is the number of bees entering and/or exiting the hive over a given period of time. Forager traffic is an important variable to monitor food availability, food demand, colony age structure, impact of pesticides, etc. on bee hives. This will lead to improved remote monitoring and general hive status and improved real time detection of the impact of pests, diseases, pesticide exposure and other hive management problems.

Semantic Segmentation processes RGB scenes and classifies pixels collectively as an object. Recent deep learning methods have shown promising results in the accuracy and the speed of semantic segmentation. However, it is inevitable for the deep learning models to fall in overfitting to data used in training due to its nature of data-centric approaches. There have been numerous Regularization methods to overcome an overfitting problem, such as data augmentation, additional loss methods such as Euclidean or Least-Square terms, and structure-related methods by adding or modifying layers like Dropout and DropConnect in a network. Among those methods, penalizing a model via an additional loss or a weight constraint does not require memory increase. With this sight, our work purposes to improve a given segmentation model through temperatures and a lightweight discriminator. Temperatures have the role of generating different versions of probability maps through the division in softmax calculations. On top of probability maps from temperatures, we concatenate a simple discriminator after the segmentation network for the competition between groundtruth feature maps and modified feature maps. We pass the additional loss calculated from those probability maps into the principal network. Our contribution consists of two parts. Firstly, we use the adversarial loss as the regularization loss in the segmentation networks and validate that it can substitute the L2 regularization loss with better validation results. Also, we apply temperatures in segmentation probability maps for providing different information without using additional convolutional layers. The experiments indicate that the spiking temperature in a generator with keeping an original probability map in a discriminator provides the model improvement in terms of pixel accuracy and mean Intersection-of-Union (mIoU). Our framework shows that the segmentation model can be improved with a small increase in training time and the number of parameters.

Indiana University-Purdue University Indianapolis (IUPUI) / Computer Vision is a mathematical tool formulated to extend human vision to machines. This tool can perform various tasks such as object classification, object tracking, motion estimation, and image segmentation. These tasks find their use in many applications, namely robotics, self-driving cars, augmented reality, and mobile applications. However, opposed to the traditional technique of incorporating handcrafted features to understand images, convolution neural networks are being used to perform the same function. Computer vision applications widely use CNNs due to their stellar performance in interpreting images. Over the years, there have been numerous advancements in machine learning, particularly to CNNs.However, the need to improve their accuracy, model size and complexity increased, making their deployment in restricted environments a challenge. Many researchers proposed techniques to reduce the size of CNN while still retaining its accuracy. Few of these include network quantization, pruning, low rank, and sparse decomposition and knowledge distillation. Some methods developed efficient models from scratch. This thesis achieves a similar goal using design space exploration techniques on the latest variant of MobileNets, MobileNet V3. Using DPD blocks, escalation in the number of expansion filters in some layers and mish activation function MobileNet V3 is reduced to 84.96% in size and made 0.2% more accurate. Furthermore, it is deployed in NXP i.MX RT1060 for image classification on CIFAR-10 dataset.