plt.semilogy(train_costs)
plt.show(block=False)
fig.savefig('./eval/' + eval_prefix + '/' + eval_prefix + '_cost.png')
fig = plt.figure()
plt.semilogy(val_errs)
plt.show(block=False)
fig.savefig('./eval/' + eval_prefix + '/' + eval_prefix + '_errs.png')
# save results
poseNet.save("./eval/{}/net_{}.pkl".format(eval_prefix, eval_prefix))
# poseNet.load("./eval/{}/net_{}.pkl".format(eval_prefix,eval_prefix))
# add prior to network
cfg = HiddenLayerParams(inputDim=(batchSize, train_gt3D_embed.shape[1]),
outputDim=(batchSize,
numpy.prod(train_gt3D.shape[1:])),
activation=None)
pcalayer = HiddenLayer(rng,
poseNet.layers[-1].output,
cfg,
copyLayer=None,
layerNum=len(poseNet.layers))
pcalayer.W.set_value(pca.components_)
pcalayer.b.set_value(pca.mean_)
poseNet.layers.append(pcalayer)
poseNet.output = pcalayer.output
poseNet.cfgParams.numJoints = train_gt3D.shape[1]
poseNet.cfgParams.nDims = train_gt3D.shape[2]
poseNet.cfgParams.outputDim = pcalayer.cfgParams.outputDim
poseNet.save("./eval/{}/network_prior.pkl".format(eval_prefix))
def __init__(self,
type=0,
nChan=1,
wIn=128,
hIn=128,
batchSize=128,
numJoints=16,
nDims=3,
resizeFactor=2):
'''
Init the parametrization
:type typeID: int
:param typeID: type of descr network
'''
super(ScaleNetParams, self).__init__()
self.batch_size = batchSize
self.numJoints = numJoints
self.nDims = nDims
if type == 1:
self.numInputs = 3
self.inpConv = 3
self.inputDim = [(batchSize, nChan, hIn, wIn),
(batchSize, nChan, hIn // resizeFactor,
wIn // resizeFactor),
(batchSize, nChan, hIn // resizeFactor**2,
wIn // resizeFactor**2)]
# Try small configuration
self.layers.append(
ConvPoolLayerParams(
inputDim=(batchSize, nChan, hIn, wIn), # w,h,nChannel
nFilters=8,
filterDim=(5, 5),
poolsize=(4, 4),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(
inputDim=(batchSize, nChan, hIn // resizeFactor,
wIn // resizeFactor), # w,h,nChannel
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(
inputDim=(batchSize, nChan, hIn // resizeFactor**2,
wIn // resizeFactor**2), # w,h,nChannel
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(5, 5),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
lout = 0
for j in range(self.numInputs):
idx = (j + 1) * self.inpConv - 1
lout += self.layers[idx].outputDim[1] * self.layers[
idx].outputDim[2] * self.layers[idx].outputDim[3]
self.layers.append(
HiddenLayerParams(inputDim=(batchSize, lout),
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(
inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, numJoints * nDims),
activation=None)) # last one is linear for regression
self.outputDim = self.layers[-1].outputDim
else:
raise NotImplementedError("not implemented")
def __init__(self, rng, inputVar=None, cfgParams=None):
"""
@see https://github.com/KaimingHe/resnet-1k-layers/blob/master/resnet-pre-act.lua
:type cfgParams: DescriptorNetParams
"""
import theano.tensor as T
self._params_filter = []
self._weights_filter = []
if cfgParams is None:
raise Exception("Cannot create a Net without config parameters (ie. cfgParams==None)")
if inputVar is None:
inputVar = T.tensor4('x') # input variable
elif isinstance(inputVar, str):
raise NotImplementedError()
self.inputVar = inputVar
self.cfgParams = cfgParams
# create network
self.layers = []
batchSize = cfgParams.batch_size
# create structure
if cfgParams.type == 0:
# Try ResNet similar configuration
depth = 47
assert (depth - 2) % 9 == 0, 'depth should be 9n+2 (e.g., 164 or 1001)'
n = (depth - 2) / 9
nStages = [32, 64, 128, 256, 256]
self.layers.append(ConvPoolLayer(rng, self.inputVar,
ConvPoolLayerParams(inputDim=self.cfgParams.inputDim, nFilters=nStages[0],
filterDim=(5, 5), stride=(1, 1),
poolsize=(2, 2), border_mode='same', activation=None,
init_method='He'),
layerNum=len(self.layers))) # one conv at the beginning
rout = self.add_res_layers(rng, self.layers[-1].output, self.layers[-1].cfgParams.outputDim, nStages[1], n, 2) # Stage 1
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[2], n, 2) # Stage 2
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[3], n, 2) # Stage 3
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[4], n, 2) # Stage 4
self.layers.append(BatchNormLayer(rng, rout, BatchNormLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(NonlinearityLayer(rng, self.layers[-1].output, NonlinearityLayerParams(inputDim=self.layers[-1].cfgParams.outputDim, activation=ReLU), layerNum=len(self.layers)))
# self.layers.append(GlobalAveragePoolLayer(rng, self.layers[-1].output, GlobalAveragePoolLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output.flatten(2),
HiddenLayerParams(# inputDim=self.layers[-1].cfgParams.outputDim,
inputDim=(self.layers[-1].cfgParams.outputDim[0], numpy.prod(self.layers[-1].cfgParams.outputDim[1:])),
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, self.cfgParams.numJoints*self.cfgParams.nDims),
activation=None),
layerNum=len(self.layers)))
self.output = self.layers[-1].output
elif cfgParams.type == 1:
# Try ResNet similar configuration
depth = 47
assert (depth - 2) % 9 == 0, 'depth should be 9n+2 (e.g., 164 or 1001)'
n = (depth - 2) / 9
nStages = [32, 64, 128, 256, 256]
self.layers.append(ConvPoolLayer(rng, self.inputVar,
ConvPoolLayerParams(inputDim=self.cfgParams.inputDim, nFilters=nStages[0],
filterDim=(5, 5), stride=(1, 1),
poolsize=(2, 2), border_mode='same', activation=None,
init_method='He'),
layerNum=len(self.layers))) # one conv at the beginning
rout = self.add_res_layers(rng, self.layers[-1].output, self.layers[-1].cfgParams.outputDim, nStages[1], n, 2) # Stage 1
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[2], n, 2) # Stage 2
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[3], n, 2) # Stage 3
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[4], n, 2) # Stage 4
self.layers.append(BatchNormLayer(rng, rout, BatchNormLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(NonlinearityLayer(rng, self.layers[-1].output, NonlinearityLayerParams(inputDim=self.layers[-1].cfgParams.outputDim, activation=ReLU), layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output.flatten(2),
HiddenLayerParams(inputDim=(self.layers[-1].cfgParams.outputDim[0], numpy.prod(self.layers[-1].cfgParams.outputDim[1:])),
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 30), activation=None),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, self.cfgParams.numJoints*self.cfgParams.nDims),
activation=None),
layerNum=len(self.layers)))
self.output = self.layers[-1].output
elif cfgParams.type == 2:
# Try ResNet similar configuration
depth = 47
assert (depth - 2) % 9 == 0, 'depth should be 9n+2 (e.g., 164 or 1001)'
n = (depth - 2) / 9
nStages = [32, 64, 128, 256, 256]
self.layers.append(ConvPoolLayer(rng, self.inputVar,
ConvPoolLayerParams(inputDim=self.cfgParams.inputDim, nFilters=nStages[0],
filterDim=(5, 5), stride=(1, 1),
poolsize=(2, 2), border_mode='same', activation=None,
init_method='He'),
layerNum=len(self.layers))) # one conv at the beginning
rout = self.add_res_layers(rng, self.layers[-1].output, self.layers[-1].cfgParams.outputDim, nStages[1], n, 2) # Stage 1
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[2], n, 2) # Stage 2
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[3], n, 2) # Stage 3
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[4], n, 2) # Stage 4
self.layers.append(BatchNormLayer(rng, rout, BatchNormLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(NonlinearityLayer(rng, self.layers[-1].output, NonlinearityLayerParams(inputDim=self.layers[-1].cfgParams.outputDim, activation=ReLU), layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output.flatten(2),
HiddenLayerParams(inputDim=(self.layers[-1].cfgParams.outputDim[0], numpy.prod(self.layers[-1].cfgParams.outputDim[1:])),
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, self.cfgParams.numJoints*self.cfgParams.nDims),
activation=None),
layerNum=len(self.layers)))
self.output = self.layers[-1].output
elif cfgParams.type == 3:
# Try ResNet similar configuration
depth = 47
assert (depth - 2) % 9 == 0, 'depth should be 9n+2 (e.g., 164 or 1001)'
n = (depth - 2) / 9
nStages = [32, 64, 128, 128, 128]
self.layers.append(ConvPoolLayer(rng, self.inputVar,
ConvPoolLayerParams(inputDim=self.cfgParams.inputDim, nFilters=nStages[0],
filterDim=(5, 5), stride=(1, 1),
poolsize=(2, 2), border_mode='same', activation=None,
init_method='He'),
layerNum=len(self.layers))) # one conv at the beginning
rout = self.add_res_layers(rng, self.layers[-1].output, self.layers[-1].cfgParams.outputDim, nStages[1], n, 2) # Stage 1
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[2], n, 2) # Stage 2
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[3], n, 2) # Stage 3
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[4], n, 2) # Stage 4
self.layers.append(BatchNormLayer(rng, rout, BatchNormLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(NonlinearityLayer(rng, self.layers[-1].output, NonlinearityLayerParams(inputDim=self.layers[-1].cfgParams.outputDim, activation=ReLU), layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output.flatten(2),
HiddenLayerParams(inputDim=(self.layers[-1].cfgParams.outputDim[0], numpy.prod(self.layers[-1].cfgParams.outputDim[1:])),
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, self.cfgParams.numJoints*self.cfgParams.nDims),
activation=None),
layerNum=len(self.layers)))
self.output = self.layers[-1].output
elif cfgParams.type == 4:
# Try ResNet similar configuration
depth = 47
assert (depth - 2) % 9 == 0, 'depth should be 9n+2 (e.g., 164 or 1001)'
n = (depth - 2) / 9
nStages = [32, 64, 128, 256, 256]
self.layers.append(ConvPoolLayer(rng, self.inputVar,
ConvPoolLayerParams(inputDim=self.cfgParams.inputDim, nFilters=nStages[0],
filterDim=(5, 5), stride=(1, 1),
poolsize=(2, 2), border_mode='same', activation=None,
init_method='He'),
layerNum=len(self.layers))) # one conv at the beginning
rout = self.add_res_layers(rng, self.layers[-1].output, self.layers[-1].cfgParams.outputDim, nStages[1], n, 2) # Stage 1
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[2], n, 2) # Stage 2
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[3], n, 2) # Stage 3
rout = self.add_res_layers(rng, rout, self.layers[-1].cfgParams.outputDim, nStages[4], n, 2) # Stage 4
self.layers.append(BatchNormLayer(rng, rout, BatchNormLayerParams(inputDim=self.layers[-1].cfgParams.outputDim), layerNum=len(self.layers)))
self.layers.append(NonlinearityLayer(rng, self.layers[-1].output, NonlinearityLayerParams(inputDim=self.layers[-1].cfgParams.outputDim, activation=ReLU), layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output.flatten(2),
HiddenLayerParams(inputDim=(self.layers[-1].cfgParams.outputDim[0], numpy.prod(self.layers[-1].cfgParams.outputDim[1:])),
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 1024), activation=ReLU),
layerNum=len(self.layers)))
self.layers.append(DropoutLayer(rng, self.layers[-1].output,
DropoutLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=self.layers[-1].cfgParams.outputDim),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, 30), activation=None),
layerNum=len(self.layers)))
self.layers.append(HiddenLayer(rng, self.layers[-1].output,
HiddenLayerParams(inputDim=self.layers[-1].cfgParams.outputDim,
outputDim=(batchSize, self.cfgParams.numJoints*self.cfgParams.nDims),
activation=None),
layerNum=len(self.layers)))
self.output = self.layers[-1].output
else:
raise NotImplementedError()
self.load(self.cfgParams.loadFile)
def __init__(self,
type=0,
n_chan=3,
w_in=128,
h_in=128,
batchSize=128,
output_dim=16):
"""
Init the parametrization
:type type: int
:param type: type of network
"""
super(PoseRegNetParams, self).__init__()
self.batch_size = batchSize
self.output_dim = output_dim
self.inputDim = (batchSize, n_chan, h_in, w_in)
if type == 0: # tiny - BB8
self.layers.append(
ConvPoolLayerParams(inputDim=(batchSize, n_chan, h_in, w_in),
nFilters=32,
filterDim=(5, 5),
poolsize=(4, 4),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=32,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=50,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
l3out = self.layers[-1].outputDim
self.layers.append(
HiddenLayerParams(inputDim=(l3out[0],
l3out[1] * l3out[2] * l3out[3]),
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, output_dim),
activation=None))
self.outputDim = self.layers[-1].outputDim
elif type == 1:
# VGG - architecure
self.layers.append(
ConvPoolLayerParams(inputDim=(batchSize, n_chan, h_in, w_in),
nFilters=64,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=64,
filterDim=(3, 3),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=128,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=128,
filterDim=(3, 3),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=256,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=256,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=256,
filterDim=(3, 3),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(2, 2),
activation=ReLU))
# Last 3 conv. layers of VGG-16 archituecture are removed from BB8,
# because of having smaller input patches
'''
self.layers.append(ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
self.layers.append(ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=512,
filterDim=(3, 3),
poolsize=(2, 2),
activation=ReLU))
'''
l3out = self.layers[-1].outputDim
self.layers.append(
HiddenLayerParams(inputDim=(l3out[0],
l3out[1] * l3out[2] * l3out[3]),
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
HiddenLayerParams(
inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, output_dim),
activation=None)) # last one is linear for regression
self.outputDim = self.layers[-1].outputDim
else:
raise NotImplementedError("not implemented")
def __init__(self,
type=0,
nChan=1,
wIn=128,
hIn=128,
batchSize=128,
numJoints=16,
nDims=3):
"""
Init the parametrization
:type type: int
:param type: type of descr network
"""
super(PoseRegNetParams, self).__init__()
self.batch_size = batchSize
self.numJoints = numJoints
self.nDims = nDims
self.inputDim = (batchSize, nChan, hIn, wIn)
if type == 0:
# Try DeepPose CNN similar configuration
self.layers.append(
ConvPoolLayerParams(
inputDim=(batchSize, nChan, hIn, wIn), # w,h,nChannel
nFilters=8,
filterDim=(5, 5),
poolsize=(4, 4),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
l3out = self.layers[-1].outputDim
self.layers.append(
HiddenLayerParams(inputDim=(l3out[0],
l3out[1] * l3out[2] * l3out[3]),
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
DropoutLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=self.layers[-1].outputDim))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
DropoutLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=self.layers[-1].outputDim))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, numJoints * nDims),
activation=None))
self.outputDim = self.layers[-1].outputDim
elif type == 11:
# Try DeepPose CNN similar configuration
self.layers.append(
ConvPoolLayerParams(
inputDim=(batchSize, nChan, hIn, wIn), # w,h,nChannel
nFilters=8,
filterDim=(5, 5),
poolsize=(4, 4),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(5, 5),
poolsize=(2, 2),
activation=ReLU))
self.layers.append(
ConvPoolLayerParams(inputDim=self.layers[-1].outputDim,
nFilters=8,
filterDim=(3, 3),
poolsize=(1, 1),
activation=ReLU))
l3out = self.layers[-1].outputDim
self.layers.append(
HiddenLayerParams(inputDim=(l3out[0],
l3out[1] * l3out[2] * l3out[3]),
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
DropoutLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=self.layers[-1].outputDim))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 1024),
activation=ReLU))
self.layers.append(
DropoutLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=self.layers[-1].outputDim))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, 30),
activation=None))
self.layers.append(
HiddenLayerParams(inputDim=self.layers[-1].outputDim,
outputDim=(batchSize, numJoints * nDims),
activation=None))
self.outputDim = self.layers[-1].outputDim
else:
raise NotImplementedError("not implemented")