In this paper, a novel approach to video temporal decomposition into semantic units, termed scenes, is presented. contrast previous segmentation approaches that employ mostly low-level visual or audiovisual features, we introduce technique jointly exploits and high-level features automatically extracted from the auditory channel. This built upon well-known method of scene transition graph (STG), first by introducing new STG approximation reduced computational cost, then extending unimodal...

This work introduces a method that combines remote sensing and deep learning into framework is tailored for accurate, reliable efficient counting sizing of plants in aerial images. The investigated task focuses on two low-density crops, potato lettuce. double objective achieved through the detection segmentation individual by fine-tuning an existing architecture called Mask R-CNN. paper includes thorough discussion optimal parametrisation to adapt R-CNN this novel task. As we examine...

Cloud masking is of central importance to the Earth Observation community. This paper deals with problem detecting clouds in visible and multispectral imagery from high-resolution satellite cameras. Recently, Machine Learning has offered promising solutions cloud masking, allowing for more flexibility than traditional thresholding techniques, which are restricted instruments requisite spectral bands. However, few studies use multi-scale features (as in, a combination pixel-level spatial)...

Digital Terrain Model (DTM) creation is essential to improving our understanding of the formation processes Martian surface. Although there have been previous demonstrations open-source or commercial planetary 3D reconstruction software, scientists are still struggling with creating good quality DTMs that meet their science needs, especially when a requirement produce large number high using “free” software. In this paper, we describe new open source system overcome many these obstacles by...

This work deals with the meta-data analysis of high-resolution orbital imagery that was acquired over last four decades Mars. The objective this is to provide a starting point for planetary scientists who are interested in examining martian surface order detect changes related not fully understood natural phenomena. An image aggregation method introduced and used generate groupings prioritising regions change detection. parameters determining each grouping season, Martian Year local time an...

Korovessis, Panagiotis MD; Sidiropoulos, Piperos, Grigoris Karagiannis, Andreas MD Author Information

In this paper, a novel approach to evaluating video temporal decomposition algorithms is presented. The evaluation measures typically used end are nonlinear combinations of precision-recall or coverage-overflow, which not metrics and additionally possess undesirable properties, such as nonsymmetricity. To alleviate these drawbacks, we introduce unidimensional measure that proven be metric satisfies number qualitative prerequisites previous do not. This named differential edit distance (DED),...

A case is presented of a motorcyclist who crashed and suffered fracture-dislocation the mid-thoracic spine without any neurologic deficit. The injury was discovered 6 weeks after crash. patient underwent surgical stabilization correction kyphotic deformity with Luque rods sublaminar wires had an uneventful recovery.

In this work the problem of automatic decomposition video into elementary semantic units, known in literature as scenes, is addressed. Two multi-modal scene segmentation techniques are proposed, both building upon Scene Transition Graph (STG). first proposed approaches, speaker diarization results used for introducing a post-processing step to STG construction algorithm, with objective discarding boundaries erroneously identified according visual-only dissimilarity. second approach, and...

Web‐based planetary image dissemination platforms usually show outline coverage of available data and offer querying for metadata as well preview download. While the usual form presenting multiorbit sets is to merge into a larger mosaic, change detection purposes it essential maintain individual images an important snapshot surface taken at specific time. We introduce web‐based interactive Mars Analysis Research System (iMARS web‐GIS), which specialized on analysis with novel tools...

In this work we deal with the problem of video concept detection, for purpose using detection results towards more effective concept-based retrieval. order to handle task, propose spatio-temporal slices, called tomographs, in same way that visual keyframes are typically used traditional keyframe-based schemes. Video tomographs capture a compact motion patterns present video, and training number base detectors. The latter augment set detectors can be trained on different image...

We address the problem of automatically coregistering planetary images to a common baseline, introducing novel generic technique that achieves an unprecedented robustness different image inputs, thus making batch-mode coregistration achievable without requiring usual parameter tweaking. introduce matching technique, which boosts performance even under most strenuous circumstances, and experimentally demonstrate validation through extensive experimental multi-instrument setup includes from...

In this paper, we address the problem of fast and accurate extraction points that correspond to same location (named tie-points) from pairs large-sized images. First, conduct a theoretical analysis performance full-image matching approach, demonstrating its limitations when applied large Subsequently, introduce novel technique impose spatial constraints on process without employing subsampled versions reference target image, which name coupled image decomposition. This splits images into...

In this paper, we deal with the problem of video concept detection to use results toward a more effective concept-based retrieval. The key novelties paper are as follows: 1) spatio-temporal slices (tomographs) in same way that visual keyframes typically used schemes. These capture compact motion patterns useful for detecting semantic concepts and training number base detectors. latter augment set keyframe-based detectors can be trained using different frame representations. 2) introduction...

This work studies the chord length distribution, in case where both ends lie on a $N$-dimensional hypersphere ($N \geq 2$). Actually, after connecting this distribution to recently estimated surface of hyperspherical cap \cite{SLi11}, closed-form expressions probability density function and cumulative are straightforwardly extracted, which followed by discussion its basic properties, among dependence from dimension. Additionally, dot product unitary vectors is estimated, problem that related length.

Abstract. The accurate split of large areas land into discrete fields is a crucial step for several agriculture-related remote sensing pipelines. This work aims to fully automate this tedious and resource-demanding process using state-of-the-art deep learning algorithm with only single Sentinel-2 image as input. Mask R-CNN, which has forged its success upon instance segmentation objects from everyday life, adapted the field boundary detection problem. Such model automatically generates...

This work examines the possibility of exploiting, for purpose video segmentation to scenes, semantic information coming from analysis visual modality. information, in contrast low-level features typically used previous approaches, is obtained by application trained concept detectors such as those developed and evaluated part TRECVID High-Level Feature Extraction Task. A large number non-binary defining a high dimensional space. In this space, each shot represented vector detector confidence...

This work presents the coregistered, orthorectified and mosaiced high-resolution products of MC11 quadrangle Mars, which have been processed using novel, fully automatic, techniques. We discuss development a pipeline that achieves automatic parameter independent geometric alignment planetary images, starting from raw input images in NASA PDS format following all required steps to produce coregistered geotiff image, corresponding footprint useful metadata. Additionally, we describe...