No abstract

In this paper, we raise important issues on scalability and the required degree of supervision of existing Mahalanobis metric learning methods. Often rather tedious optimization procedures are applied that become computationally intractable on a large scale. Further, if one considers the constantly growing amount of data it is often infeasible to specify fully supervised labels for all data points. Instead, it is easier to specify labels in form of equivalence constraints. We introduce a simple though effective strategy to learn a distance metric from equivalence constraints, based on a statistical inference perspective. In contrast to existing methods we do not rely on complex optimization problems requiring computationally expensive iterations. Hence, our method is orders of magnitudes faster than comparable methods. Results on a variety of challenging benchmarks with rather diverse nature demonstrate the power of our method. These include faces in unconstrained environments, matching before unseen object instances and person re-identification across spatially disjoint cameras. In the latter two benchmarks we clearly outperform the state-ofthe-art.

Very recently tracking was approached using classification techniques such as support vector machines. The object to be tracked is discriminated by a classifier from the background. In a similar spirit we propose a novel on-line AdaBoost feature selection algorithm for tracking. The distinct advantage of our method is its capability of on-line training. This allows to adapt the classifier while tracking the object. Therefore appearance changes of the object (e.g. out of plane rotations, illumination changes) are handled quite naturally. Moreover, depending on the background the algorithm selects the most discriminating features for tracking resulting in stable tracking results. By using fast computable features (e.g. Haar-like wavelets, orientation histograms, local binary patterns) the algorithm runs in real-time. We demonstrate the performance of the algorithm on several (publically available) video sequences.

The Photodetector Array Camera and Spectrometer (PACS) is one of the three science instruments on ESA's far infrared and submillimetre observatory. It employs two Ge:Ga photoconductor arrays (stressed and unstressed) with 16 × 25 pixels, each, and two filled silicon bolometer arrays with 16 × 32 and 32 × 64 pixels, respectively, to perform integral-field spectroscopy and imaging photometry in the 60−210 μm wavelength regime. In photometry mode, it simultaneously images two bands, 60−85 μm or 85−125 μm and 125−210 μm, over a field of view of ∼1.75 × 3.5 , with close to Nyquist beam sampling in each band. In spectroscopy mode, it images a field of 47 × 47 , resolved into 5 × 5 pixels, with an instantaneous spectral coverage of ∼ 1500 km s −1 and a spectral resolution of ∼175 km s −1 . We summarise the design of the instrument, describe observing modes, calibration, and data analysis methods, and present our current assessment of the in-orbit performance of the instrument based on the performance verification tests. PACS is fully operational, and the achieved performance is close to or better than the pre-launch predictions. Key words. space vehicles: instruments -instrumentation: photometers -instrumentation: spectrographsHerschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA.

No abstract