Searcharxiv⌕ Search

arXiv subjects

Shabab Bazrafkan

Publications and source records attributed to Shabab Bazrafkan.

13 recordsLinked to original sources
arXiv2022

PolyWorld: Polygonal Building Extraction with Graph Neural Networks in Satellite Images

While most state-of-the-art instance segmentation methods produce binary segmentation masks, geographic and cartographic applications typically require precise vector polygons of extracted objects instead of rasterized output. This paper introduces PolyWorld, a neural network that directly extracts building vertices from an image and connects them correctly to create precise polygons. The model predicts the connection strength between each pair of vertices using a graph neural network and estimates the assignments by solving a differentiable optimal transport problem. Moreover, the vertex positions are optimized by minimizing a combined segmentation and polygonal angle difference loss. PolyWorld significantly outperforms the state of the art in building polygonization and achieves not only notable quantitative results, but also produces visually pleasing building polygons. Code and trained weights are publicly available at https://github.com/zorzi-s/PolyWorldPretrainedNetwork.

cs.CV↗
arXiv2020

Re-Training StyleGAN -- A First Step Towards Building Large, Scalable Synthetic Facial Datasets

StyleGAN is a state-of-art generative adversarial network architecture that generates random 2D high-quality synthetic facial data samples. In this paper, we recap the StyleGAN architecture and training methodology and present our experiences of retraining it on a number of alternative public datasets. Practical issues and challenges arising from the retraining process are discussed. Tests and validation results are presented and a comparative analysis of several different re-trained StyleGAN weightings is provided 1. The role of this tool in building large, scalable datasets of synthetic facial data is also discussed.

cs.NE↗
arXiv2019

To Recurse or not to Recurse,a Low Dose CT Study

Restoring high-quality CT images from low dose CT counterparts is an ill-posed, nonlinear problem to which Deep Learning approaches have been giving superior solutions compared to classical model-based approaches. In this article, a framework is presented wherein a Recurrent Neural Network (RNN) is utilized to remove the streaking artefacts from low projection number CT imaging. The results indicate similar image restoration performance for the RNN compared to the feedforward network in low noise cases while in high noise levels the RNN returns better results. The computational costs are also compared between RNN and feedforward networks.

eess.IV↗
arXiv2019

Deep Neural Network Assisted Iterative Reconstruction Method for Low Dose CT

Low Dose Computed Tomography suffers from a high amount of noise and/or undersampling artefacts in the reconstructed image. In the current article, a Deep Learning technique is exploited as a regularization term for the iterative reconstruction method SIRT. While SIRT minimizes the error in the sinogram space, the proposed regularization model additionally steers intermediate SIRT reconstructions towards the desired output. Extensive evaluations demonstrate the superior outcomes of the proposed method compared to the state of the art techniques. Comparing the forward projection of the reconstructed image with the original signal shows a higher fidelity to the sinogram space for the current approach amongst other learning based methods.

eess.IV↗
arXiv2019

Deep Neural Network and Data Augmentation Methodology for off-axis iris segmentation in wearable headsets

A data augmentation methodology is presented and applied to generate a large dataset of off-axis iris regions and train a low-complexity deep neural network. Although of low complexity the resulting network achieves a high level of accuracy in iris region segmentation for challenging off-axis eye-patches. Interestingly, this network is also shown to achieve high levels of performance for regular, frontal, segmentation of iris regions, comparing favorably with state-of-the-art techniques of significantly higher complexity. Due to its lower complexity, this network is well suited for deployment in embedded applications such as augmented and mixed reality headsets.

cs.CV↗
arXiv2018

Versatile Auxiliary Classifier with Generative Adversarial Network (VAC+GAN), Multi Class Scenarios

Conditional generators learn the data distribution for each class in a multi-class scenario and generate samples for a specific class given the right input from the latent space. In this work, a method known as "Versatile Auxiliary Classifier with Generative Adversarial Network" for multi-class scenarios is presented. In this technique, the Generative Adversarial Networks (GAN)'s generator is turned into a conditional generator by placing a multi-class classifier in parallel with the discriminator network and backpropagate the classification error through the generator. This technique is versatile enough to be applied to any GAN implementation. The results on two databases and comparisons with other method are provided as well.

cs.LG↗
arXiv2018

Versatile Auxiliary Classifier with Generative Adversarial Network (VAC+GAN)

One of the most interesting challenges in Artificial Intelligence is to train conditional generators which are able to provide labeled adversarial samples drawn from a specific distribution. In this work, a new framework is presented to train a deep conditional generator by placing a classifier in parallel with the discriminator and back propagate the classification error through the generator network. The method is versatile and is applicable to any variations of Generative Adversarial Network (GAN) implementation, and also gives superior results compared to similar methods.

eess.IV↗
arXiv2018

Versatile Auxiliary Regressor with Generative Adversarial network (VAR+GAN)

Being able to generate constrained samples is one of the most appealing applications of the deep generators. Conditional generators are one of the successful implementations of such models wherein the created samples are constrained to a specific class. In this work, the application of these networks is extended to regression problems wherein the conditional generator is restrained to any continuous aspect of the data. A new loss function is presented for the regression network and also implementations for generating faces with any particular set of landmarks is provided.

eess.IV↗
arXiv2018

Face Synthesis with Landmark Points from Generative Adversarial Networks and Inverse Latent Space Mapping

Facial landmarks refer to the localization of fundamental facial points on face images. There have been a tremendous amount of attempts to detect these points from facial images however, there has never been an attempt to synthesize a random face and generate its corresponding facial landmarks. This paper presents a framework for augmenting a dataset in a latent Z-space and applied to the regression problem of generating a corresponding set of landmarks from a 2D facial dataset. The BEGAN framework has been used to train a face generator from CelebA database. The inverse of the generator is implemented using an Adam optimizer to generate the latent vector corresponding to each facial image, and a lightweight deep neural network is trained to map latent Z-space vectors to the landmark space. Initial results are promising and provide a generic methodology to augment annotated image datasets with additional intermediate samples.

eess.IV↗
arXiv2017

An End to End Deep Neural Network for Iris Segmentation in Unconstraint Scenarios

With the increasing imaging and processing capabilities of today's mobile devices, user authentication using iris biometrics has become feasible. However, as the acquisition conditions become more unconstrained and as image quality is typically lower than dedicated iris acquisition systems, the accurate segmentation of iris regions is crucial for these devices. In this work, an end to end Fully Convolutional Deep Neural Network (FCDNN) design is proposed to perform the iris segmentation task for lower-quality iris images. The network design process is explained in detail, and the resulting network is trained and tuned using several large public iris datasets. A set of methods to generate and augment suitable lower quality iris images from the high-quality public databases are provided. The network is trained on Near InfraRed (NIR) images initially and later tuned on additional datasets derived from visible images. Comprehensive inter-database comparisons are provided together with results from a selection of experiments detailing the effects of different tunings of the network. Finally, the proposed model is compared with SegNet-basic, and a near-optimal tuning of the network is compared to a selection of other state-of-art iris segmentation algorithms. The results show very promising performance from the optimized Deep Neural Networks design when compared with state-of-art techniques applied to the same lower quality datasets.

eess.IV↗
arXiv2017

Smart Augmentation - Learning an Optimal Data Augmentation Strategy

A recurring problem faced when training neural networks is that there is typically not enough data to maximize the generalization capability of deep neural networks(DNN). There are many techniques to address this, including data augmentation, dropout, and transfer learning. In this paper, we introduce an additional method which we call Smart Augmentation and we show how to use it to increase the accuracy and reduce overfitting on a target network. Smart Augmentation works by creating a network that learns how to generate augmented data during the training process of a target network in a way that reduces that networks loss. This allows us to learn augmentations that minimize the error of that network. Smart Augmentation has shown the potential to increase accuracy by demonstrably significant measures on all datasets tested. In addition, it has shown potential to achieve similar or improved performance levels with significantly smaller network sizes in a number of tested cases.

cs.AI↗