Multi-shot person re-identification via relational stein divergence

Person re-identification is particularly challenging due to significant appearance changes across separate camera views. In order to re-identify people, a representative human signature should effectively handle differences in illumination, pose and camera parameters. While general appearance-based methods are modelled in Euclidean spaces, it has been argued that some applications in image and video analysis are better modelled via non-Euclidean manifold geometry. To this end, recent approaches represent images as covariance matrices, and interpret such matrices as points on Riemannian manifolds. As direct classification on such manifolds can be difficult, in this paper we propose to represent each manifold point as a vector of similarities to class representers, via a recently introduced form of Bregman matrix divergence known as the Stein divergence. This is followed by using a discriminative mapping of similarity vectors for final classification. The use of similarity vectors is in contrast to the traditional approach of embedding manifolds into tangent spaces, which can suffer from representing the manifold structure inaccurately. Comparative evaluations on benchmark ETHZ and iLIDS datasets for the person re-identification task show that the proposed approach obtains better performance than recent techniques such as Histogram Plus Epitome, Partial Least Squares, and Symmetry-Driven Accumulation of Local Features.

Improved image set classification via joint sparse approximated nearest subspaces

Existing multi-model approaches for image set classification extract local models by clustering each image set individually only once, with fixed clusters used for matching with other image sets. However, this may result in the two closest clusters to represent different characteristics of an object, due to different undesirable environmental conditions (such as variations in illumination and pose). To address this problem, we propose to constrain the clustering of each query image set by forcing the clusters to have resemblance to the clusters in the gallery image sets. We first define a Frobenius norm distance between subspaces over Grassmann manifolds based on reconstruction error. We then extract local linear subspaces from a gallery image set via sparse representation. For each local linear subspace, we adaptively construct the corresponding closest subspace from the samples of a probe image set by joint sparse representation. We show that by minimising the sparse representation reconstruction error, we approach the nearest point on a Grassmann manifold. Experiments on Honda, ETH-80 and Cambridge-Gesture datasets show that the proposed method consistently outperforms several other recent techniques, such as Affine Hull based Image Set Distance (AHISD), Sparse Approximated Nearest Points (SANP) and Manifold Discriminant Analysis (MDA).

Second-harmonic generation from biological tissues : effect of excitation wavelength

We report on the measurement of second-harmonic signals from hyperplastic parenchyma and stroma in malignant human prostate tissue under femtosecond pulsed illumination in the wavelength range from 730 to 870 nm. In particular, the relationship of the second-harmonic generation to the excitation wavelength is measured. The result in these two regions behaves considerably differently and thus provides a possible indicator for identifying tissue components and malignancy.

All-environment visual place recognition with SMART

This paper presents Sequence Matching Across Route Traversals (SMART); a generally applicable sequence-based place recognition algorithm. SMART provides invariance to changes in illumination and vehicle speed while also providing moderate pose invariance and robustness to environmental aliasing. We evaluate SMART on vehicles travelling at highly variable speeds in two challenging environments; firstly, on an all-terrain vehicle in an off-road, forest track and secondly, using a passenger car traversing an urban environment across day and night. We provide comparative results to the current state-of-the-art SeqSLAM algorithm and investigate the effects of altering SMART’s image matching parameters. Additionally, we conduct an extensive study of the relationship between image sequence length and SMART’s matching performance. Our results show viable place recognition performance in both environments with short 10-metre sequences, and up to 96% recall at 100% precision across extreme day-night cycles when longer image sequences are used.

Visible-light-induced selective photocatalytic oxidation of benzylamine into imine over supported Ag/AgI photocatalysts

Titanate nanotubes (TNT) supported AgI nanoparticles were prepared by a two-step method: the deposition of Ag2O on titanate nanotubes from AgNO3 solution and the subsequent I-adsorption process from NaI solution. It is found that the supported AgI samples exhibited excellent photoactivity for the selective oxidation of benzylamine to the corresponding imine under visible light illumination and the photocatalyst can be used for many times without apparent activity loss. X-ray diffraction studies, transmission electron microscopy, diffuse reflectance UV-Vis spectroscopy and nitrogen adsorption measurements were used for the characterization of the as-prepared and recycled AgI samples. It is found that under visible light irradiation, AgI partially decomposed to produce Ag/AgI nanostructure and thus stabilized. The photoactivity of supported Ag/AgI for the selective oxidation of benzylamine was studied in terms of the light intensity, wavelength, temperature and substituent. It is proposed that the formation of plasmonic Ag nanoparticles should be responsible for the high activity and selectivity.

Synthesis, thin-film morphology, and comparative study of bulk and bilayer heterojunction organic photovoltaic devices using soluble diketopyrrolopyrrole molecules

Diketopyrrolopyrrole (DPP)-based organic semiconductors EH-DPP-TFP and EH-DPP-TFPV with branched ethyl-hexyl solubilizing alkyl chains and end capped with trifluoromethyl phenyl groups were designed and synthesized via Suzuki coupling. These compounds show intense absorptions up to 700 nm, and thin film-forming characteristics that sensitively depend on the solvent and coating conditions. Both materials have been used as electron donors in bulk heterojunction and bilayer organic photovoltaic (OPV) devices with fullerenes as acceptors and their performance has been studied in detail. The best power conversion efficiency of 3.3% under AM1.5G illumination (100 mW cm -2) was achieved for bilayer solar cells when EH-DPP-TFPV was used with C 60, after a thermal annealing step to induce dye aggregation and interdiffusion of C 60 with the donor material. To date, this is one of the highest efficiencies reported for simple bilayer OPV devices.

Poly(2,5-bis(2-octyldodecyl)-3,6-di(furan-2-yl)-2,5-dihydro-pyrrolo[3,4-c] pyrrole-1,4-dione-co-thieno[3,2-b]thiophene): A high performance polymer semiconductor for both organic thin film transistors and organic photovoltaics

A new diketopyrrolopyrrole (DPP)-containing donor-acceptor polymer, poly(2,5-bis(2-octyldodecyl)-3,6-di(furan-2-yl)-2,5-dihydro-pyrrolo[3,4-c] pyrrole-1,4-dione-co-thieno[3,2-b]thiophene) (PDBF-co-TT), is synthesized and studied as a semiconductor in organic thin film transistors (OTFTs) and organic photovoltaics (OPVs). High hole mobility of up to 0.53 cm 2 V -1 s -1 in bottom-gate, top-contact OTFT devices is achieved owing to the ordered polymer chain packing and favoured chain orientation, strong intermolecular interactions, as well as uniform film morphology of PDBF-co-TT. The optimum band gap of 1.39 eV and high hole mobility make this polymer a promising donor semiconductor for the solar cell application. When paired with a fullerene acceptor, PC 71BM, the resulting OPV devices show a high power conversion efficiency of up to 4.38% under simulated standard AM1.5 solar illumination.

Thiophene-benzothiadiazole-thiophene (D-A-D) based polymers : Effect of donor/acceptor moieties adjacent to D-A-D segment on photophysical and photovoltaic properties

New push-pull copolymers based on thiophene (donor) and benzothiadiazole (acceptor) units, poly[4,7-bis(3-dodecylthiophene-2-yl) benzothiadiazole-co- thiophene] (PT3B1) and poly[4,7-bis(3-dodecylthiophene-2-yl) benzothiadiazole-co-benzothiadiazole] (PT2B2), are designed and synthesized via Stille and Suzuki coupling routes respectively. Gel permeation chromatography shows the number average molecular weights are 31100 and 8400 g mol-1 for the two polymers, respectively. Both polymers have shown absorption throughout a wide range of the UV-vis region, from 300 to 650 nm. A significant red shift of the absorption edge is observed in thin films compared to solution of the copolymers; the optical band gap is in the range of 1.7 to 1.8 eV. Cyclic voltammetry indicates reversible oxidation and reduction processes with HOMO energy levels calculated to be in the range of 5.2 to 5.4 eV. Upon testing both materials for organic field-effect transistors (OFETs), PT3B1 showed a hole mobility of 6.1 × 10-4 cm2 V-1 s -1, while PT2B2 did not show any field effect transport. Both copolymers displayed a photovoltaic response when combined with a methanofullerene as an electron acceptor. The best performance was achieved when the copolymer PT3B1 was blended with [70]PCBM in a 1:4 ratio, exhibiting a short-circuit current of 7.27 mA cm-2, an open circuit voltage of 0.85 V, and a fill factor of 41% yielding a power conversion efficiency of 2.54% under simulated air mass (AM) 1.5 global (1.5 G) illumination conditions (100 mW cm-2). Similar devices utilizing PT2B2 in place of PT3B1 demonstrated reduced performance with a short-circuit current of 4.8 mA cm -2, an open circuit voltage of 0.73 V, and a fill factor of 30% resulting in a power conversion efficiency of roughly 1.06%.

Ultrathin hematite films deposited layer-by-layer on TiO2 underlayer for efficient water splitting under visible light

Ultrathin hematite (α-Fe2O3) film deposited on a TiO2 underlayer as a photoanode for photoelectrochemical water splitting was described. The TiO2 underlayer was coated on conductive fluorine-doped tin oxide (FTO) glass by spin coating. The hematite films were formed layer-by-layer by repeating the separated two-phase hydrolysis-solvothermal reaction of iron(III) acetylacetonate and aqueous ammonia. A photocurrent density of 0.683 mA cm−2 at +1.5 V vs. RHE (reversible hydrogen electrode) was obtained under visible light (>420 nm, 100 mW cm−2) illumination. The TiO2 underlayer plays an important role in the formation of hematite film, acting as an intermediary to alleviate the dead layer effect and as a support of large surface areas to coat greater amounts of Fe2O3. The as-prepared photoanodes are notably stable and highly efficient for photoelectrochemical water splitting under visible light. This study provides a facile synthesis process for the controlled production of highly active ultrathin hematite film and a simple route for photocurrent enhancement using several photoanodes in tandem.

Size dependent photodegradation of CdS particles deposited onto TiO2 mesoporous films by SILAR method

The properties of CdS nanoparticles incorporated onto mesoporous TiO2 films by a successive ionic layer adsorption and reaction (SILAR) method were investigated by Raman spectroscopy, UV-visible spectroscopy, transmission electron microscopy (TEM) and X-ray photoelectron spectroscopy (XPS). High resolution TEM indicated that the synthesized CdS particles were hexagonal phase and the particle sizes were less than 5 nm when SILAR cycles were fewer than 9. Quantum size effect was found with the CdS sensitized TiO2 films prepared with up to 9 SILAR cycles. The band gap of CdS nanoparticles decreased from 2.65 eV to 2.37 eV with the increase of the SILAR cycles from 1 to 11. The investigation of the stability of the CdS/TiO2 films in air under illumination (440.6 µW/cm2) showed that the photodegradation rate was up to 85% per day for the sample prepared with 3 SILAR cycles. XPS analysis indicated that the photodegradation was due to the oxidation of CdS, leading to the transformation from sulphide to sulphate (CdSO4). Furthermore, the degradation rate was strongly dependent upon the particle size of CdS. Smaller particles showed faster degradation rate. The size-dependent photo-induced oxidization was rationalized with the variation of size-dependent distribution of surface atoms of CdS particles. Molecular Dynamics (MD) simulation has indicated that the surface sulphide anion of a large CdS particle such as CdS made with 11 cycles (CdS11, particle size = 5.6 nm) accounts for 9.6% of the material whereas this value is increased to 19.2% for (CdS3) based smaller particles (particle size: 2.7 nm). Nevertheless, CdS nanoparticles coated with ZnS material showed a significantly enhanced stability under illumination in air. A nearly 100% protection of CdS from photon induced oxidation with a ZnS coating layer prepared using four SILAR cycles, suggesting the formation of a nearly complete coating layer on the CdS nanoparticles.

Fourier active appearance models

Gaining invariance to camera and illumination variations has been a well investigated topic in Active Appearance Model (AAM) fitting literature. The major problem lies in the inability of the appearance parameters of the AAM to generalize to unseen conditions. An attractive approach for gaining invariance is to fit an AAM to a multiple filter response (e.g. Gabor) representation of the input image. Naively applying this concept with a traditional AAM is computationally prohibitive, especially as the number of filter responses increase. In this paper, we present a computationally efficient AAM fitting algorithm based on the Lucas-Kanade (LK) algorithm posed in the Fourier domain that affords invariance to both expression and illumination. We refer to this as a Fourier AAM (FAAM), and show that this method gives substantial improvement in person specific AAM fitting performance over traditional AAM fitting methods.

Lighting invariant urban street classification

In this paper we propose the hybrid use of illuminant invariant and RGB images to perform image classification of urban scenes despite challenging variation in lighting conditions. Coping with lighting change (and the shadows thereby invoked) is a non-negotiable requirement for long term autonomy using vision. One aspect of this is the ability to reliably classify scene components in the presence of marked and often sudden changes in lighting. This is the focus of this paper. Posed with the task of classifying all parts in a scene from a full colour image, we propose that lighting invariant transforms can reduce the variability of the scene, resulting in a more reliable classification. We leverage the ideas of “data transfer” for classification, beginning with full colour images for obtaining candidate scene-level matches using global image descriptors. This is commonly followed by superpixellevel matching with local features. However, we show that if the RGB images are subjected to an illuminant invariant transform before computing the superpixel-level features, classification is significantly more robust to scene illumination effects. The approach is evaluated using three datasets. The first being our own dataset and the second being the KITTI dataset using manually generated ground truth for quantitative analysis. We qualitatively evaluate the method on a third custom dataset over a 750m trajectory.

Grateful fateful sunshine rain

Grateful Fateful Sunshine Rain is a permanent public artwork commissioned by Aria Property Group through a competitive process for the Austin apartment building in South Brisbane. Artist Statement: Residents of Brisbane have a complex relationship with weather. As the capital of the Sunshine State, weather is an integral part of the city’s cultural identity. Weather deeply affects the mood of the city – from the excitement of scantily clad partygoers on balmy December evenings and late February’s lethargy, to the deepening anxiety that emerges after 100 days of rain (or more commonly, 100 days without rain). With a brief nod to the city’s – now decommissioned – iconic MCL weather beacon, Grateful Fateful Sunshine Rain taps into this aspect of Brisbane’s psyche with poetic, illuminated visualisations of real-time weather forecasts issued by the Bureau of Meteorology. Each evening, the artwork downloads tomorrow’s forecast from the Bureau of Meteorology website. Data including, current local temperature, humidity, wind speed & direction, precipitation (rain, hail etc), are used to generate a lighting display that conveys how tomorrow will feel. The artwork’s background colour indicates the expected temperature – from cold blues through mild pastel pinks and blues to bright hot oranges and reds. White fluffy clouds roll across the artwork if cloud is predicted. The density of these clouds indicates the level of cover whilst movement indicates expected wind speed and direction. If rain is predicted, sparkles of white light will appear on top of whichever background colour is chosen for the next day’s temperature. Sparkles appear constantly before wet, drizzly days, and intermittently if scattered showers are predicted. Intermittent, but more intense sparkles appear before rain storms or thunderstorms. Research Contribution: The work has made contributions to the field in the way it rethinks approaches to the conceptualization, design and realization of illuminated urban media. This has led to new theorizations of urban media, which consider light and illumination can be used to convey meaningful data. The research has produced new methods for controlling illumination systems using tools and techniques typically employed in computation arts. It has also develop methods and processes for the design and production of illuminated urban media architectures that are connected to real time data sources, and do which not follow the assumed logics of screen based media and displays.

Representing team behaviours from noisy data using player role

Due to their unobtrusive nature, vision-based approaches to tracking sports players have been preferred over wearable sensors as they do not require the players to be instrumented for each match. Unfortunately however, due to the heavy occlusion between players, variation in resolution and pose, in addition to fluctuating illumination conditions, tracking players continuously is still an unsolved vision problem. For tasks like clustering and retrieval, having noisy data (i.e. missing and false player detections) is problematic as it generates discontinuities in the input data stream. One method of circumventing this issue is to use an occupancy map, where the field is discretised into a series of zones and a count of player detections in each zone is obtained. A series of frames can then be concatenated to represent a set-play or example of team behaviour. A problem with this approach though is that the compressibility is low (i.e. the variability in the feature space is incredibly high). In this paper, we propose the use of a bilinear spatiotemporal basis model using a role representation to clean-up the noisy detections which operates in a low-dimensional space. To evaluate our approach, we used a fully instrumented field-hockey pitch with 8 fixed high-definition (HD) cameras and evaluated our approach on approximately 200,000 frames of data from a state-of-the-art real-time player detector and compare it to manually labeled data.

Temporal characteristics of melanopsin inputs to the human pupil light reflex

Rods, cones and melanopsin containing intrinsically photosensitive retinal ganglion cells (ipRGCs) operate in concert to regulate pupil diameter. The temporal properties of intrinsic ipRGC signalling are distinct to those of rods and cones, including longer latencies and sustained signalling after light offset. We examined whether the melanopsin mediated post-illumination pupil response (PIPR) and pupil constriction were dependent upon the inter-stimulus interval (ISI) between successive light pulses and the temporal frequency of sinusoidal light stimuli. Melanopsin excitation was altered by variation of stimulus wavelength (464 nm and 638 nm lights) and irradiance (11.4 and 15.2 log photons cm(-2) s(-1)). We found that 6s PIPR amplitude was independent of ISI and temporal frequency for all melanopsin excitation levels, indicating complete summation. In contrast to the PIPR, the maximum pupil constriction increased with increasing ISI with high and low melanopsin excitation, but time to minimum diameter was slower with high melanopsin excitation only. This melanopsin response to briefly presented pulses (16 and 100 ms) slows the temporal response of the maximum pupil constriction. We also demonstrate that high melanopsin excitation attenuates the phasic peak-trough pupil amplitude compared to conditions with low melanopsin excitation, indicating an interaction between inner and outer retinal inputs to the pupil light reflex. We infer that outer retina summation is important for rapidly controlling pupil diameter in response to short timescale fluctuations in illumination and may occur at two potential sites, one that is presynaptic to extrinsic photoreceptor input to ipRGCs, or another within the pupil control pathway if ipRGCs have differential temporal tuning to extrinsic and intrinsic signalling.