Removing noise in digital images is a fundamental operation that arises in many application domains. In this paper we consider the median filter, a filtering technique that replaces the color of each pixel with the median of those in a square neighborhood of fixed radius. For some use cases, the size of the neighborhood or the image depth may be large, making existing algorithms either too slow, or not applicable at all due to excessive memory requirements. In this paper we describe architecture-specific optimizations that enable the computation of the median filter with arbitrary window size and image depth on multicore processors and GPUs. We report preliminary results that indicate that the parallel implementations are suitable for practical use, with the GPU version outperforming the CPU.

MOTIVATION: In fluorescence microscopy, single-molecule localization microscopy (SMLM) techniques aim at localizing with high-precision high-density fluorescent molecules by stochastically activating and imaging small subsets of blinking emitters. Super resolution plays an important role in this field since it allows to go beyond the intrinsic light diffraction limit. RESULTS: In this work, we propose a deep learning-based algorithm for precise molecule localization of high-density frames acquired by SMLM techniques whose ℓ2-based loss function is regularized by non-negative and ℓ0-based constraints. The ℓ0 is relaxed through its continuous exact ℓ0 (CEL0) counterpart. The arising approach, named DeepCEL0, is parameter-free, more flexible, faster and provides more precise molecule localization maps if compared to the other state-of-the-art methods. We validate our approach on both simulated and real fluorescence microscopy data. AVAILABILITY AND IMPLEMENTATION: DeepCEL0 code is freely accessible at https://github.com/sedaboni/DeepCEL0.

We here propose to collect and analyze large amounts of multimedia data from different public and private sources, in the form of text, image, video, to predict relevant information about specific quantities related to fashion brands, such as their sales volumes and/or trends. To this aim, we deem deep learning techniques as the suitable instrument capable of managing extremely large amounts of multimedia data. While a few works exist in literature on learning applications in the fashion area, where text is used to perform sentiment analysis operations, limited research has also considered images and videos in this context. In this paper, starting with an overview of the state of the art of the applications of artificial intelligence for fashion, we set the stage for a holistic approach for the deep learning based analysis of multimedia data related to fashion.