In this supplementary material, we make use of the following notation. x i denotes the i th entry of the (column) vector X(x), y j the j th entry of the (column) vector Y (y), V[i; j] denotes the entry in position (i, j) of the matrix V. Also, V[ ; j] denotes the j th column of the matrix V. Finally, δ i,j denotes the delta function which gives 1 if i = j, and 0 otherwise. 7. Example of a distr...

We present an image-based method for separating diffuse and specular reflections using environmental structured illumination. Two types of structured illumination are discussed: phase-shifted sine wave patterns, and phaseshifted binary stripe patterns. In both cases the low-pass filtering nature of diffuse reflections is utilized to separate the reflection components. We illustrate our method o...

We first provide our derivation of the coordinate-ascent method used in the alternating inference, which searches for the best scoring x and v. The overview of our method is given in Algorithm 1 in the main paper. We now show how to rewrite the scoring function S(x,v) as its quadratic form w.r.t. v. Note that γ(v(y)) = (1− δ)v(y) + δ. So we can write the first term (i.e., the global mask term) ...

001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050 051 052 053 054 055 056 057 058 059 060 061 062 063 064 065 066 067 068 069 070 071 072 073 074 075 076 077 078 079 080 081 082 083 084 085 086 087 088 089 090 091 092 093 094 095 096 097 098 099 1...

This section provides the detailed proof of the equivalence between pattern potentials and RBMs. The high level idea of the proof is to treat each hidden variable in an RBM as encoding a pattern. We first introduce the definition of pattern potentials by Rother et al. in [2], a few necessary change of variable tricks, and two different ways to compose more general high order potentials, “sum” a...

S1 . Parameter estimation for finetuning with structured SVM objective . . . . . . . . . . . . . . . . . . . . . 1 S2 . Details on hard negative data mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 S3 . Implementation details on model parameter estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 S4 . Efficiency of fine-grained search (FGS) ....

The polarization state of an ultrafast laser is dynamically controlled using two Spatial Light Modulators and additional waveplates. Consequently, four states of polarization, linear horizontal and vertical, radial and azimuthal, all with a ring intensity distribution, were dynamically switched at a frequency ν = 12.5 Hz while synchronized with a motion control system. This technique, demonstra...

Using learning material in an adaptive way requires a preparation beforehand. In order to guide the process of creating learning material, criteria for adaptation and techniques for content preparation have to be known. The aim of this paper is to give first hints for such guidance. Therefore, adaptation criteria and provisions which can be used to realize such a preparation should be identifie...

The increase in centrosome size in mitosis was described over a century ago, and yet it is poorly understood how centrioles, which lie at the core of centrosomes, organize the pericentriolar material (PCM) in this process. Now, structured illumination microscopy reveals in Drosophila that, before clouds of PCM appear, its proteins are closely associated with interphase centrioles in two tube-li...