|
|
Efficient implementation of deep neural networks
Kopál, Jakub ; Mrázová, Iveta (advisor) ; Střelský, Jakub (referee)
In recent years, algorithms in the area of object detection have constantly been improving. The success of these algorithms has reached a level, where much of the development is focused on increasing speed at the expense of accuracy. As a result of recent improvements in the area of deep learning and new hardware architectures optimized for deep learning models, it is possible to detect objects in an image several hundreds times per second using only embedded and mobile devices. The main objective of this thesis is to study and summarize the most important methods in the area of effective object detection and apply them to a given real-world problem. By using state-of- the-art methods, we developed a traction-by-detection algorithm, which is based on our own object detection models that track transport vehicles in real-time using embedded and mobile devices. 1
|
|
|
Optical Music Recognition using Deep Neural Networks
Mayer, Jiří ; Pecina, Pavel (advisor) ; Hajič, Jan (referee)
Optical music recognition is a challenging field similar in many ways to optical text recognition. It brings, however, many challenges that traditional pipeline-based recog- nition systems struggle with. The end-to-end approach has proven to be superior in the domain of handwritten text recognition. We tried to apply this approach to the field of OMR. Specifically, we focused on handwritten music recognition. To resolve the lack of training data, we developed an engraving system for handwritten music called Mashcima. This engraving system is successful at mimicking the style of the CVC- MUSCIMA dataset. We evaluated our model on a portion of the CVC-MUSCIMA dataset and the approach seems to be promising. 1
|
|
|
Semi-Supervised Training of Deep Neural Networks for Speech Recognition
Veselý, Karel ; Ircing, Pavel (referee) ; Lamel, Lori (referee) ; Burget, Lukáš (advisor)
V této dizertační práci nejprve prezentujeme teorii trénování neuronových sítí pro rozpoznávání řeči společně s implementací trénovacího receptu 'nnet1', který je součástí toolkitu s otevřeným kódem Kaldi. Recept se skládá z předtrénování bez učitele pomocí algoritmu RBM, trénování klasifikátoru z řečových rámců s kriteriální funkcí Cross-entropy a ze sekvenčního trénování po větách s kriteriální funkcí sMBR. Následuje hlavní téma práce, kterým je semi-supervised trénování se smíšenými daty s přepisem i bez přepisu. Inspirováni konferenčními články a úvodními experimenty jsme se zaměřili na několik otázek: Nejprve na to, zda je lepší konfidence (t.j. důvěryhodnosti automaticky získaných anotací) počítat po větách, po slovech nebo po řečových rámcích. Dále na to, zda by konfidence měly být použity pro výběr dat nebo váhování dat - oba přístupy jsou kompatibilní s trénováním pomocí metody stochastického nejstrmějšího sestupu, kde jsou gradienty řečových rámců násobeny vahou. Dále jsme se zabývali vylepšováním semi-supervised trénování pomocí kalibrace kofidencí a přístupy, jak model dále vylepšit pomocí dat se správným přepisem. Nakonec jsme navrhli jednoduchý recept, pro který není nutné časově náročné ladění hyper-parametrů trénování, a který je prakticky využitelný pro různé datové sady. Experimenty probíhaly na několika sadách řečových dat: pro rozpoznávač vietnamštiny s 10 přepsaným hodinami (Babel) se chybovost snížila o 2.5%, pro angličtinu se 14 přepsanými hodinami (Switchboard) se chybovost snížila o 3.2%. Zjistili jsme, že je poměrně těžké dále vylepšit přesnost systému pomocí úprav konfidencí, zároveň jsme ale přesvědčení, že naše závěry mají značnou praktickou hodnotu: data bez přepisu je jednoduché nasbírat a naše navrhované řešení přináší dobrá zlepšení úspěšnosti a není těžké je replikovat.
|
|
|
Exploring Benefits of Transfer Learning in Neural Machine Translation
Kocmi, Tom ; Bojar, Ondřej (advisor) ; van Genabith, Josef (referee) ; Cuřin, Jan (referee)
Title: Exploring Benefits of Transfer Learning in Neural Machine Translation Author: Tom Kocmi Department: Institute of Formal and Applied Linguistics Supervisor: doc. RNDr. Ondřej Bojar, Ph.D., Institute of Formal and Applied Linguistics Keywords: transfer learning, machine translation, deep neural networks, low-resource languages Abstract: Neural machine translation is known to require large numbers of parallel train- ing sentences, which generally prevent it from excelling on low-resource lan- guage pairs. This thesis explores the use of cross-lingual transfer learning on neural networks as a way of solving the problem with the lack of resources. We propose several transfer learning approaches to reuse a model pretrained on a high-resource language pair. We pay particular attention to the simplicity of the techniques. We study two scenarios: (a) when we reuse the high-resource model without any prior modifications to its training process and (b) when we can prepare the first-stage high-resource model for transfer learning in advance. For the former scenario, we present a proof-of-concept method by reusing a model trained by other researchers. In the latter scenario, we present a method which reaches even larger improvements in translation performance. Apart from proposed techniques, we focus on an...
|
|
|
Comparison of deep learning and classical methods for traffic signs detection
Geiger, Petr ; Šikudová, Elena (advisor) ; Mirbauer, Martin (referee)
The goal of this thesis is to explore and evaluate classic and deep neural network computer vision methods in the task of detection position of a level crossing barrier. This thesis is based on an initial detection algorithm using a Stable Wave Detector. The initial algorithm is optimized both in performance and quality of the results. Both is crucial, because the best method should be suitable as a component of the real-time level crossing safety system. Then an another approach is implemented using deep neural networks and optimized in the same manner. Throughout the work several datasets are created for both training and testing of the algorithms. Both approaches are finally evaluated on the same test datasets and the results are compared.
|
|
|
Indonesian-English Neural Machine Translation
Dwiastuti, Meisyarah ; Popel, Martin (advisor) ; Novák, Michal (referee)
Title: Indonesian-English Neural Machine Translation Author: Meisyarah Dwiastuti Department: Institute of Formal and Applied Linguistics Supervisor: Mgr. Martin Popel, Ph.D., Institute of Formal and Applied Linguis- tics Abstract: In this thesis, we conduct a study on neural machine translation (NMT) for an under-studied language, Indonesian, specifically for English-Indonesian (EN-ID) and Indonesian-English (ID-EN) in a low-resource domain, TED talks. Our goal is to implement domain adaptation methods to improve the low-resource EN-ID and ID-EN NMT systems. First, we implement model fine-tuning method for EN-ID and ID-EN NMT systems by leveraging a large parallel corpus contain- ing movie subtitles. Our analysis shows the benefit of this method for the improve- ment of both systems. Second, we improve our ID-EN NMT system by leveraging English monolingual corpora through back-translation. Our back-translation ex- periments focus on how to incorporate the back-translated monolingual corpora to the training set, in which we investigate various existing training regimes and introduce a novel 4-way-concat training regime. We also analyze the effect of fine- tuning our back-translation models with different scenarios. Experimental results show that our method of implementing back-translation followed by model...
|
|
|
Deep Neural Networks Approximation
Stodůlka, Martin ; Mrázek, Vojtěch (referee) ; Vaverka, Filip (advisor)
The goal of this work is to find out the impact of approximated computing on accuracy of deep neural network, specifically neural networks for image classification. A version of framework Caffe called Ristretto-caffe was chosen for neural network implementation, which was extended for the use of approximated operations. Approximated computing was used for multiplication in forward pass for convolution. Approximated components from Evoapproxlib were chosen for this work.
|
|
|
Pedestrian Identification
Jurča, Jan ; Špaňhel, Jakub (referee) ; Hradiš, Michal (advisor)
This thesis deals with pedestrian identification from video sequence based on person, face and gait recognition. For person and face recognition are used pretrained networks. While for gait recognition is implemented and compared many different networks. Final pedestrian recognition is based on multimodal fusion realized by neural network. For the purpose of the work was created dataset, along with a set of tools that allow its almost automatic creation.
|
|
|
Convolutional Networks for Document Layout Analysis
Endrych, David ; Herout, Adam (referee) ; Kodym, Oldřich (advisor)
The goal of this thesis is to create a tool for analyzig the page layouts of text documents. The problem is solved by convolution neural networks. The architecture chosen in this thesis is the U-Net architecture. The cross entropy error function with weight map is used for train the network model. Paragraph regions are obtained throught connected component analysis. Experiments are evaluated using the Symmetric Best Dice object metric. Experiments have shown that it is better to use all paragraph edges than to focus only on vertical paragraph edges. In addition, experiments show that batche sampling strategies and adaptive resolution help to improve analysis results. The experiments also describe the application of separators, which is useful in analyzing multi-column documents.
|
|
|
Deep Neural Networks for Defect Detection
Juřica, Tomáš ; Herout, Adam (referee) ; Hradiš, Michal (advisor)
The goal of this work is to bring automatic defect detection to the manufacturing process of plastic cards. A card is considered defective when it is contaminated with a dust particle or a hair. The main challenges I am facing to accomplish this task are a very few training data samples (214 images), small area of target defects in context of an entire card (average defect area is 0.0068 \% of the card) and also very complex background the detection task is performed on. In order to accomplish the task, I decided to use Mask R-CNN detection algorithm combined with augmentation techniques such as synthetic dataset generation. I trained the model on the synthetic dataset consisting of 20 000 images. This way I was able to create a model performing 0.83 AP at 0.1 IoU on the original data test set.
|
guest ::
