These days deep learning is the fastest-growing field in the field of Machine Learning (ML) and Deep Neural Networks (DNN). Among many of DNN structures, the Convolutional Neural Networks (CNN) are currently the main tool used for the image analysis and classification purposes. Although great achievements and perspectives, deep neural networks and accompanying learning algorithms have some relevant challenges to tackle. In this paper, we have focused on the most frequently mentioned problem in the field of machine learning, that is the lack of sufficient amount of the training data or uneven class balance within the datasets. One of the ways of dealing with this problem is so called data augmentation. In the paper we have compared and analyzed multiple methods of data augmentation in the task of image classification, starting from classical image transformations like rotating, cropping, zooming, histogram based methods and finishing at Style Transfer and Generative Adversarial Networks, along with the representative examples. Next, we presented our own method of data augmentation based on image style transfer. The method allows to generate the new images of high perceptual quality that combine the content of a base image with the appearance of another ones. The newly created images can be used to pre-train the given neural network in order to improve the training process efficiency. Proposed method is validated on the three medical case studies: skin melanomas diagnosis, histopathological images and breast magnetic resonance imaging (MRI) scans analysis, utilizing the image classification in order to provide a diagnose. In such kind of problems the data deficiency is one of the most relevant issues. Finally, we discuss the advantages and disadvantages of the methods being analyzed.

Waste pollution is one of the most significant environmental issues in the modern world. The importance of recycling is well known, both for economic and ecological reasons, and the industry demands high efficiency. Current studies towards automatic waste detection are hardly comparable due to the lack of benchmarks and widely accepted standards regarding the used metrics and data. Those problems are addressed in this article by providing a critical analysis of over ten existing waste datasets and a brief but constructive review of the existing Deep Learning-based waste detection approaches. This article collects and summarizes previous studies and provides the results of authors' experiments on the presented datasets, all intended to create a first replicable baseline for litter detection. Moreover, new benchmark datasets detect-waste and classify-waste are proposed that are merged collections from the above-mentioned open-source datasets with unified annotations covering all possible waste categories: bio, glass, metal and plastic, non-recyclable, other, paper, and unknown. Finally, a two-stage detector for litter localization and classification is presented. EfficientDet-D2 is used to localize litter, and EfficientNet-B2 to classify the detected waste into seven categories. The classifier is trained in a semi-supervised fashion making the use of unlabeled images. The proposed approach achieves up to 70% of average precision in waste detection and around 75% of classification accuracy on the test dataset. The code and annotations used in the studies are publicly available online¹.

Deep neural networks have achieved great success in many domains. However, successful deployment of such systems is determined by proper manual selection of the neural architecture. This is a tedious and time-consuming process that requires expert knowledge. Different tasks need very different architectures to obtain satisfactory results. The group of methods called the neural architecture search (NAS) helps to find effective architecture in an automated manner. In this paper, we present the use of an architecture search framework to solve the medical task of malignant melanoma detection. Unlike many other methods tested on benchmark datasets, we tested it on practical problem, which differs greatly in terms of difficulty in distinguishing between classes, resolution of images, data balance within the classes, and the number of data available. In order to find a suitable network structure, the hill-climbing search strategy was employed along with network morphism operations to explore the search space. The network morphism operations allow for incremental increases in the network size with the use of the previously trained network. This kind of knowledge reusing allows significantly reducing the computational cost. The proposed approach produces structures that achieve similar results to those provided by manually designed structures, at the same time making use of almost 20 times fewer parameters. What is more, the search process lasts on average only 18h on single GPU.

The paper presents the results of research on the use of Deep Neural Networks (DNN) for automatic classification of the skin lesions. The authors have focused on the most effective kind of DNNs for image processing, namely Convolutional Neural Networks (CNN). In particular, three kinds of CNN were analyzed: VGG19, Residual Networks (ResNet) and the hybrid of VGG19 CNN with the Support Vector Machine (SVM). The research was carried out with the use of database of over 10 000 images representing skin lesions: benign and malignant. Because of an uneven number of images representing different classes of lesions, the up-sampling of underrepresented class was applied. The comparison of the CNN structures with respect to the accuracy, sensitivity and specificity was performed using k-fold validation method.

To successfully train a deep neural network, a large amount of human-labeled data is required. Unfortunately, in many areas, collecting and labeling data is a difficult and tedious task. Several ways have been developed to mitigate the problem associated with the shortage of data, the most common of which is transfer learning. However, in many cases, the use of transfer learning as the only remedy is insufficient. In this study, we improve deep neural models training and increase the classification accuracy under a scarcity of data by the use of the self-supervised learning technique. Self-supervised learning allows an unlabeled dataset to be used for pretraining the network, as opposed to transfer learning that requires labeled datasets. The pretrained network can be then fine-tuned using the annotated data. Moreover, we investigated the effect of combining the self-supervised learning approach with transfer learning. It is shown that this strategy outperforms network training from scratch or with transfer learning. The tests were conducted on a very important and sensitive application (skin lesion classification), but the presented approach can be applied to a broader family of applications, especially in the medical domain where the scarcity of data is a real problem.