def test_gibbs(self):
infname = 'gibbs'
param = opengm.inferenceParameter(self.gm, alg=infname)
param.set(steps=2000, burnInSteps=20,
variableProposal=opengm.VariableProposal.cyclic)
param.isAcyclic = opengm.Tribool.maybe
self.runAlgTester(infname, parameter=param)
def test_astar(self):
infname='astar'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(heuristic=opengm.Heuristic.standard,maxHeapSize=2000000,numberOfOpt=1)
self.runAlgTester(infname,optimal=True,parameter=param)
param.heuristic=opengm.Heuristic.fast
self.runAlgTester(infname,optimal=True,parameter=param)
param.heuristic=opengm.Heuristic.default
self.runAlgTester(infname,optimal=True,parameter=param)
for x4 in range(2):
labelsum=x1+x2+x3+x4
if labelsum is not 2 and labelsum is not 0 :
fClosedness[x1,x2,x3,x4]=high
fidClosedness=gm.addFunction(fClosedness)
# for each boundary in the grid, i.e. for each variable
# of the model, add one 1st order functions
# and one 1st order factor
# and for each junction of four inter-pixel edges on the grid,
# one factor is added that connects the corresponding variable
# indices and refers to the closedness-function
for yt in range(tGridSize[1]):
for xt in range(tGridSize[0]):
# unaries
if (xt % 2 + yt % 2) == 1 :
gradient = abs( data[xt / 2, yt / 2]- data[xt/ 2 + xt % 2, yt / 2 + yt % 2])
f=numpy.array([beta*gradient , (1.0-beta)*(1.0-gradient)],dtype=numpy.float32)
print gm[gm.addFactor(gm.addFunction(f),[cToVi.convert(xt,yt)])].asNumpy()
# high order factors (4.th order)
if xt % 2 + yt % 2 == 2 :
vi=[cToVi.convert(xt + 1, yt),cToVi.convert(xt - 1, yt),cToVi.convert(xt, yt + 1), cToVi.convert(xt, yt - 1)]
vi=sorted(vi);
gm.addFactor(fidClosedness,vi)
param=opengm.inferenceParameter(gm=gm,alg="lf")
param.set(maxSubgraphSize=4)
inf=opengm.inferenceAlgorithm(gm=gm,alg="lf",parameter=param)
inf.infer(verbose=True)
arg=inf.arg()
printSolution(data,arg)
def test_bp(self):
infname='libdai-bp'
param=opengm.inferenceParameter(self.gm,alg=infname)
self.runAlgTester(infname,parameter=param)
def runAlgTester(self,algName,optimal=False,parameter=None):
if parameter is None:
parameter=opengm.inferenceParameter(self.gm,alg=algName)
inf=opengm.inferenceAlgorithm(self.gm,alg=algName,parameter=parameter)
checkInference(self.gm,inf,self.argOpt,optimal=optimal)
def test_lazyflipper(self):
param=opengm.inferenceParameter(self.gm,alg='lf')
param.maxSubgraphSize=2
self.runAlgTester('lf',parameter=param)
param.maxSubgraphSize=3
self.runAlgTester('lf',parameter=param)
def test_icm(self):
param=opengm.inferenceParameter(self.gm,alg='icm')
param.moveType=opengm.MoveType.variable
self.runAlgTester('icm',parameter=param)
param.moveType=opengm.MoveType.factor
self.runAlgTester('icm',parameter=param)
def test_trbp(self):
infname='trbp'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(isAcyclic=opengm.Tribool.false,steps=100)
param.isAcyclic=opengm.Tribool.maybe
self.runAlgTester(infname,parameter=param)
def test_bp(self):
infname='bp'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(isAcyclic=opengm.Tribool.false)
param.isAcyclic=opengm.Tribool.false
self.runAlgTester(infname,parameter=param)
def test_cplex(self):
infname='lpcplex'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(integerConstraint=True)
self.runAlgTester(infname,parameter=param)
def test_junction_tree(self):
infname='libdai-jt'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set()
self.runAlgTester(infname,parameter=param)
param.set(updateRule=opengm.JunctionTreeUpdateRule.shsh,heuristic=opengm.JunctionTreeHeuristic.minfill)
def test_trbp(self):
infname='libdai-trbp'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(steps=100,updateRule=opengm.BpUpdateRule.parall,tolerance=0.0001,damping=0.01,ntrees=0)
self.runAlgTester(infname,parameter=param)
def test_gibbs(self):
infname='gibbs'
param=opengm.inferenceParameter(self.gm,alg=infname)
param.set(steps=2000)
param.isAcyclic=opengm.Tribool.maybe
self.runAlgTester(infname,parameter=param)
print "get regions pixels"
regionPixels=[ ]
regionFeatures=numpy.ones([rag.numberOfRegions(),3],dtype=numpy.float64)
for r in range(rag.numberOfRegions()):
regionPixels.append(rag.regionPixels(r))
print "build model"
gm=buildGM(img=img,rag=rag,dataImage=seedMapTemp2,numLabels=numLabels,boundaryPixels=boundaryPixels,regionPixels=regionPixels,beta=beta,sigma=sigma,verbose=False)
print "start inference"
alg='trbp'
param=opengm.inferenceParameter(gm=gm,alg=alg)
param.set(steps=6,damping=0.1)
#arg=inferAndWriteResult(printNth=1,alg=alg,param=param,img=imgRGB,rag=rag,prefix=outprefix,filename=basename,boundaryPixels=boundaryPixels,regionPixels=regionPixels,verbose=verboseInf)
alg='bp'
param=opengm.inferenceParameter(gm=gm,alg=alg)
param.set(steps=1,damping=0.6)
arg=inferAndWriteResult(printNth=1,alg=alg,param=param,img=imgRGB,rag=rag,prefix=outprefix,filename=basename,boundaryPixels=boundaryPixels,regionPixels=regionPixels,verbose=verboseInf)
alg='icm'
param=opengm.inferenceParameter(gm=gm,alg=alg)
param.set()
#arg=inferAndWriteResult(printNth=1000,alg=alg,param=param,img=imgRGB,rag=rag,prefix=outprefix,filename=basename,boundaryPixels=boundaryPixels,regionPixels=regionPixels,verbose=verboseInf,startingPoint=arg)
alg='gibbs'
param=opengm.inferenceParameter(gm=gm,alg=alg)
