Data-Augmentation for Reducing Dataset Bias in Person Re-identification

In this paper we explore ways to address the issue of dataset bias in person re-identification by using data augmentation to increase the variability of the available datasets, and we introduce a novel data augmentation method for re-identification based on changing the image background. We show that use of data augmentation can improve the cross-dataset generalisation of convolutional network based re-identification systems, and that changing the image background yields further improvements.

Die Balanced Scorecard als Führungsinstrument zur Herbarbeleg-Digitalisierung

Neben circa 22.000 verschiedenen kultivierten Pflanzenarten besitzt die Zentraleinrichtung Botanischer Garten und Botanische Museum Berlin-Dahlem (BGBM) ca. 3,8 Millionen konservierter Pflanzenbelege aus aller Welt in seinem Herbarium. Diese Herbarbelege dienen als Grundlage wissenschaftlicher Forschung. Über die Erstellung digitaler Kopien kann das Versenden des empfindlichen Materials verhindert und der Zugang für Forscher erheblich erleichtert werden. Die Prozesse und Methoden der Digitalisierung werden in Herbar Digital untersucht, um eine Senkung der Kosten zu ermöglichen. In dem vorliegenden Arbeitspapier wird der Einsatz einer Balanced Scorecard als Steuerungsinstrument der Herbarbeleg-Digitalisierung untersucht. Das Instrument misst die Zielerreichung und unterstützt so Entscheidungen der Führung. Vision und Strategie der Unternehmung werden in mehreren Perspektiven operationalisiert. Für jedes Ziel sind Indikatoren, Sollwerte, Istwerte und Maßnahmen zu definieren. Die Istwerte werden zu den Indikatoren und Zeitpunkten verdichtet und zeigen so den Zielerreichungsgrad an. In einem individuellen Dashboard werden die Ziele, Indikatoren, Zielwerte, Maßnahmen und weitere Informationen angezeigt und können kommentiert werden. Die Planwerte sind die Basis für den operativen Planungsprozess. Die Struktur sieht vor, dass die Ziel-Dashboards aller Organisationseinheiten einer Hochschule schließlich zu Sub- und Teil-Balanced Scorecards miteinander verknüpft werden können. Es ergibt sich je nach Sachverhalt eine Übersicht über die Ziele und korrespondierenden Indikatoren sowie deren Entwicklung und Maßnahmen. Dies wird am Beispiel der Herbarbeleg-Digitalisierung verdeutlicht.

Die Prozesskostenrechnung zum Kosten- und Leistungsmanagement der Herbarbeleg-Digitalisierung

Die Prozesskostenrechnung untersucht die Verteilung der Vollkosten auf die bei der Herbarbeleg-Digitalisierung anfallenden Tätigkeiten (hier exemplarisch in 2010). Außerdem stellt dieses Modell ein Instrument zum Monitoring neu eingesetzter Produkt- und Prozessinnovationen dar. Die Tätigkeiten sind in Projektkostenstellen abgegrenzt, um z. B. nach Einzel- und Gemeinkosten zu unterscheiden. Der Fokus liegt auf der Methodenentwicklung. Die Führungsaufgaben und unterstützenden Tätigkeiten sind in Bezug auf die Leistungsmenge neutral. Ihre Kosten werden auf leistungsmengeninduzierende Kerntätigkeiten umgelegt, die in sieben Phasenbereiche konsolidiert wurden. Für die untersuchten Projekte „Willdenow“ und „Global Plant Initiative“ ergaben sich jeweils Leistungs- und Kostenkennzahlen für die einzelnen Phasenbereiche. Entwicklungsbedarf des Modells besteht u. a. bei der Validität der Kostenträger, Zeichenerfassung, Kosten ehrenamtlicher Tätigkeiten, Reduktion der Komplexität und sonstiger Störungen. Die Prozesskostenrechnung ist als ein weiterführendes Instrument der Vollkostenrechnung entwickelt worden und kann auch als Deckungsbeitragsrechnung betrieben werden. Mit den Ergebnissen wird keine Kontrolle der Aufgabenwahrnehmung einzelner Mitarbeiter verfolgt.

ExtSym: a program to aid space-group determination from powder diffraction data

Once unit-cell dimensions have been determined from a powder diffraction data set and therefore the crystal system is known (e.g. orthorhombic), the method presented by Markvardsen, David, Johnson & Shankland [Acta Cryst. (2001), A57, 47-54] can be used to generate a table ranking the extinction symbols of the given crystal system according to probability. Markvardsen et al. tested a computer program (ExtSym) implementing the method against Pawley refinement outputs generated using the TF12LS program [David, Ibberson & Matthewman (1992). Report RAL-92-032. Rutherford Appleton Laboratory, Chilton, Didcot, Oxon, UK]. Here, it is shown that ExtSym can be used successfully with many well known powder diffraction analysis packages, namely DASH [David, Shankland, van de Streek, Pidcock, Motherwell & Cole (2006). J. Appl. Cryst. 39, 910-915], FullProf [Rodriguez-Carvajal (1993). Physica B, 192, 55-69], GSAS [Larson & Von Dreele (1994). Report LAUR 86-748. Los Alamos National Laboratory, New Mexico, USA], PRODD [Wright (2004). Z. Kristallogr. 219, 1-11] and TOPAS [Coelho (2003). Bruker AXS GmbH, Karlsruhe, Germany]. In addition, a precise description of the optimal input for ExtSym is given to enable other software packages to interface with ExtSym and to allow the improvement/modification of existing interfacing scripts. ExtSym takes as input the powder data in the form of integrated intensities and error estimates for these intensities. The output returned by ExtSym is demonstrated to be strongly dependent on the accuracy of these error estimates and the reason for this is explained. ExtSym is tested against a wide range of data sets, confirming the algorithm to be very successful at ranking the published extinction symbol as the most likely. (C) 2008 International Union of Crystallography Printed in Singapore - all rights reserved.

A semiotic framework for information representation of energy consumption in office buildings

Taking a perspective from a whole building lifecycle, occupier's actions could account for about 50% of energy. However occupants' activities influence building energy performance is still a blind area. Building energy performance is thought to be the result of a combination of building fabrics, building services and occupants' activities, along with their interactions. In this sense, energy consumption in built environment is regarded as a socio-technical system. In order to understand how such a system works, a range of physical, technical and social information is involved that needs to be integrated and aligned. This paper has proposed a semiotic framework to add value for Building Information Modelling, incorporating energy-related occupancy factors in a context of office buildings. Further, building information has been addressed semantically to describe a building space from the facility management perspective. Finally, the framework guides to set up building information representation system, which can help facility managers to manage buildings efficiently by improving their understanding on how office buildings are operated and used.

An annotation-free method for evaluating privacy protection techniques in videos

While several privacy protection techniques are pre- sented in the literature, they are not complemented with an established objective evaluation method for their assess- ment and comparison. This paper proposes an annotation- free evaluation method that assesses the two key aspects of privacy protection that are privacy and utility. Unlike some existing methods, the proposed method does not rely on the use of subjective judgements and does not assume a spe- cific target type in the image data. The privacy aspect is quantified as an appearance similarity and the utility aspect is measured as a structural similarity between the original raw image data and the privacy-protected image data. We performed an extensive experimentation using six challeng- ing datasets (including two new ones) to demonstrate the effectiveness of the evaluation method by providing a per- formance comparison of four state-of-the-art privacy pro- tection techniques.

Temporally stable feature clusters for maritime object tracking in visible and thermal imagery

This paper describes a new approach to detect and track maritime objects in real time. The approach particularly addresses the highly dynamic maritime environment, panning cameras, target scale changes, and operates on both visible and thermal imagery. Object detection is based on agglomerative clustering of temporally stable features. Object extents are first determined based on persistence of detected features and their relative separation and motion attributes. An explicit cluster merging and splitting process handles object creation and separation. Stable object clus- ters are tracked frame-to-frame. The effectiveness of the approach is demonstrated on four challenging real-world public datasets.

Tracking performance evaluation on PETS 2015 Challenge datasets

This paper presents a quantitative evaluation of a tracking system on PETS 2015 Challenge datasets using well-established performance measures. Using the existing tools, the tracking system implements an end-to-end pipeline that include object detection, tracking and post- processing stages. The evaluation results are presented on the provided sequences of both ARENA and P5 datasets of PETS 2015 Challenge. The results show an encouraging performance of the tracker in terms of accuracy but a greater tendency of being prone to cardinality error and ID changes on both datasets. Moreover, the analysis show a better performance of the tracker on visible imagery than on thermal imagery.

Vehicle subtype, make and model classification from side profile video

This paper addresses the challenging domain of vehicle classification from pole-mounted roadway cameras, specifically from side-profile views. A new public vehicle dataset is made available consisting of over 10000 side profile images (86 make/model and 9 sub-type classes). 5 state-of-the-art classifiers are applied to the dataset, with the best achieving high classification rates of 98.7% for sub-type and 99.7- 99.9% for make and model recognition, confirming the assertion made that single vehicle side profile images can be used for robust classification.

Meeting detection in video through semantic analysis

In this paper we present a novel approach to detect people meeting. The proposed approach works by translating people behaviour from trajectory information into semantic terms. Having available a semantic model of the meeting behaviour, the event detection is performed in the semantic domain. The model is learnt employing a soft-computing clustering algorithm that combines trajectory information and motion semantic terms. A stable representation can be obtained from a series of examples. Results obtained on a series of videos with different types of meeting situations show that the proposed approach can learn a generic model that can effectively be applied on the behaviour recognition of meeting situations.