from __future__ import absolute_import from __future__ import division from __future__ import print_function import mxnet as mx import numpy as np from config import config ACT_BIT = 1 bn_mom = 0.9 workspace = 256 memonger = False def Conv(**kwargs): body = mx.sym.Convolution(**kwargs) return body def Act(data, act_type, name): if act_type=='prelu': body = mx.sym.LeakyReLU(data = data, act_type='prelu', name = name) else: body = mx.symbol.Activation(data=data, act_type=act_type, name=name) return body #def lin(data, num_filter, workspace, name, binarize, dcn): # bit = 1 # if not binarize: # if not dcn: # conv1 = Conv(data=data, num_filter=num_filter, kernel=(1,1), stride=(1,1), pad=(0,0), # no_bias=True, workspace=workspace, name=name + '_conv') # bn1 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn') # act1 = Act(data=bn1, act_type='relu', name=name + '_relu') # return act1 # else: # bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn') # act1 = Act(data=bn1, act_type='relu', name=name + '_relu') # conv1_offset = mx.symbol.Convolution(name=name+'_conv_offset', data = act1, # num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) # conv1 = mx.contrib.symbol.DeformableConvolution(name=name+"_conv", data=act1, offset=conv1_offset, # num_filter=num_filter, pad=(1,1), kernel=(3, 3), num_deformable_group=1, stride=(1, 1), dilate=(1, 1), no_bias=False) # #conv1 = Conv(data=act1, num_filter=num_filter, kernel=(3,3), stride=(1,1), pad=(1,1), # # no_bias=False, workspace=workspace, name=name + '_conv') # return conv1 # else: # bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn') # act1 = Act(data=bn1, act_type='relu', name=name + '_relu') # conv1 = mx.sym.QConvolution_v1(data=act1, num_filter=num_filter, kernel=(1,1), stride=(1,1), pad=(0,0), # no_bias=True, workspace=workspace, name=name + '_conv', act_bit=ACT_BIT, weight_bit=bit) # conv1 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn2') # return conv1 def lin3(data, num_filter, workspace, name, k, g=1, d=1): if k!=3: conv1 = Conv(data=data, num_filter=num_filter, kernel=(k,k), stride=(1,1), pad=((k-1)//2,(k-1)//2), num_group=g, no_bias=True, workspace=workspace, name=name + '_conv') else: conv1 = Conv(data=data, num_filter=num_filter, kernel=(k,k), stride=(1,1), pad=(d,d), num_group=g, dilate=(d, d), no_bias=True, workspace=workspace, name=name + '_conv') bn1 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name + '_bn') act1 = Act(data=bn1, act_type='relu', name=name + '_relu') ret = act1 return ret def ConvFactory(data, num_filter, kernel, stride=(1, 1), pad=(0, 0), act_type="relu", mirror_attr={}, with_act=True, dcn=False, name=''): if not dcn: conv = mx.symbol.Convolution( data=data, num_filter=num_filter, kernel=kernel, stride=stride, pad=pad, no_bias=True, workspace=workspace, name=name+'_conv') else: conv_offset = mx.symbol.Convolution(name=name+'_conv_offset', data = data, num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) conv = mx.contrib.symbol.DeformableConvolution(name=name+"_conv", data=data, offset=conv_offset, num_filter=num_filter, pad=(1,1), kernel=(3,3), num_deformable_group=1, stride=stride, dilate=(1, 1), no_bias=False) bn = mx.symbol.BatchNorm(data=conv, fix_gamma=False, momentum=bn_mom, eps=2e-5, name=name+'_bn') if with_act: act = Act(bn, act_type, name=name+'_relu') #act = mx.symbol.Activation( # data=bn, act_type=act_type, attr=mirror_attr, name=name+'_relu') return act else: return bn class CAB: def __init__(self, data, nFilters, nModules, n, workspace, name, dilate, group): self.data = data self.nFilters = nFilters self.nModules = nModules self.n = n self.workspace = workspace self.name = name self.dilate = dilate self.group = group self.sym_map = {} def get_output(self, w, h): key = (w, h) if key in self.sym_map: return self.sym_map[key] ret = None if h==self.n: if w==self.n: ret = (self.data, self.nFilters) else: x = self.get_output(w+1, h) f = int(x[1]*0.5) if w!=self.n-1: body = lin3(x[0], f, self.workspace, "%s_w%d_h%d_1"%(self.name, w, h), 3, self.group, 1) else: body = lin3(x[0], f, self.workspace, "%s_w%d_h%d_1"%(self.name, w, h), 3, self.group, self.dilate) ret = (body,f) else: x = self.get_output(w+1, h+1) y = self.get_output(w, h+1) if h%2==1 and h!=w: xbody = lin3(x[0], x[1], self.workspace, "%s_w%d_h%d_2"%(self.name, w, h), 3, x[1]) #xbody = xbody+x[0] else: xbody = x[0] #xbody = x[0] #xbody = lin3(x[0], x[1], self.workspace, "%s_w%d_h%d_2"%(self.name, w, h), 3, x[1]) if w==0: ybody = lin3(y[0], y[1], self.workspace, "%s_w%d_h%d_3"%(self.name, w, h), 3, self.group) else: ybody = y[0] ybody = mx.sym.concat(y[0], ybody, dim=1) body = mx.sym.add_n(xbody,ybody, name="%s_w%d_h%d_add"%(self.name, w, h)) body = body/2 ret = (body, x[1]) self.sym_map[key] = ret return ret def get(self): return self.get_output(1, 1)[0] def conv_resnet(data, num_filter, stride, dim_match, name, binarize, dcn, dilate, **kwargs): bit = 1 #print('in unit2') # the same as https://github.com/facebook/fb.resnet.torch#notes, a bit difference with origin paper bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn1') if not binarize: act1 = Act(data=bn1, act_type='relu', name=name + '_relu1') conv1 = Conv(data=act1, num_filter=int(num_filter*0.5), kernel=(1,1), stride=(1,1), pad=(0,0), no_bias=True, workspace=workspace, name=name + '_conv1') else: act1 = mx.sym.QActivation(data=bn1, act_bit=ACT_BIT, name=name + '_relu1', backward_only=True) conv1 = mx.sym.QConvolution(data=act1, num_filter=int(num_filter*0.5), kernel=(1,1), stride=(1,1), pad=(0,0), no_bias=True, workspace=workspace, name=name + '_conv1', act_bit=ACT_BIT, weight_bit=bit) bn2 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn2') if not binarize: act2 = Act(data=bn2, act_type='relu', name=name + '_relu2') conv2 = Conv(data=act2, num_filter=int(num_filter*0.5), kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, workspace=workspace, name=name + '_conv2') else: act2 = mx.sym.QActivation(data=bn2, act_bit=ACT_BIT, name=name + '_relu2', backward_only=True) conv2 = mx.sym.QConvolution(data=act2, num_filter=int(num_filter*0.5), kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, workspace=workspace, name=name + '_conv2', act_bit=ACT_BIT, weight_bit=bit) bn3 = mx.sym.BatchNorm(data=conv2, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn3') if not binarize: act3 = Act(data=bn3, act_type='relu', name=name + '_relu3') conv3 = Conv(data=act3, num_filter=num_filter, kernel=(1,1), stride=(1,1), pad=(0,0), no_bias=True, workspace=workspace, name=name + '_conv3') else: act3 = mx.sym.QActivation(data=bn3, act_bit=ACT_BIT, name=name + '_relu3', backward_only=True) conv3 = mx.sym.QConvolution(data=act3, num_filter=num_filter, kernel=(1,1), stride=(1,1), pad=(0,0), no_bias=True, workspace=workspace, name=name + '_conv3', act_bit=ACT_BIT, weight_bit=bit) #if binarize: # conv3 = mx.sym.BatchNorm(data=conv3, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn4') if dim_match: shortcut = data else: if not binarize: shortcut = Conv(data=act1, num_filter=num_filter, kernel=(1,1), stride=stride, no_bias=True, workspace=workspace, name=name+'_sc') else: shortcut = mx.sym.QConvolution(data=act1, num_filter=num_filter, kernel=(1,1), stride=stride, pad=(0,0), no_bias=True, workspace=workspace, name=name + '_sc', act_bit=ACT_BIT, weight_bit=bit) if memonger: shortcut._set_attr(mirror_stage='True') return conv3 + shortcut def conv_hpm(data, num_filter, stride, dim_match, name, binarize, dcn, dilation, **kwargs): bit = 1 #print('in unit2') # the same as https://github.com/facebook/fb.resnet.torch#notes, a bit difference with origin paper bn1 = mx.sym.BatchNorm(data=data, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn1') if not binarize: act1 = Act(data=bn1, act_type='relu', name=name + '_relu1') if not dcn: conv1 = Conv(data=act1, num_filter=int(num_filter*0.5), kernel=(3,3), stride=(1,1), pad=(dilation,dilation), dilate=(dilation,dilation), no_bias=True, workspace=workspace, name=name + '_conv1') else: conv1_offset = mx.symbol.Convolution(name=name+'_conv1_offset', data = act1, num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) conv1 = mx.contrib.symbol.DeformableConvolution(name=name+'_conv1', data=act1, offset=conv1_offset, num_filter=int(num_filter*0.5), pad=(1,1), kernel=(3, 3), num_deformable_group=1, stride=(1, 1), dilate=(1, 1), no_bias=True) else: act1 = mx.sym.QActivation(data=bn1, act_bit=ACT_BIT, name=name + '_relu1', backward_only=True) conv1 = mx.sym.QConvolution_v1(data=act1, num_filter=int(num_filter*0.5), kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, workspace=workspace, name=name + '_conv1', act_bit=ACT_BIT, weight_bit=bit) bn2 = mx.sym.BatchNorm(data=conv1, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn2') if not binarize: act2 = Act(data=bn2, act_type='relu', name=name + '_relu2') if not dcn: conv2 = Conv(data=act2, num_filter=int(num_filter*0.25), kernel=(3,3), stride=(1,1), pad=(dilation,dilation), dilate=(dilation,dilation), no_bias=True, workspace=workspace, name=name + '_conv2') else: conv2_offset = mx.symbol.Convolution(name=name+'_conv2_offset', data = act2, num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) conv2 = mx.contrib.symbol.DeformableConvolution(name=name+'_conv2', data=act2, offset=conv2_offset, num_filter=int(num_filter*0.25), pad=(1,1), kernel=(3, 3), num_deformable_group=1, stride=(1, 1), dilate=(1, 1), no_bias=True) else: act2 = mx.sym.QActivation(data=bn2, act_bit=ACT_BIT, name=name + '_relu2', backward_only=True) conv2 = mx.sym.QConvolution_v1(data=act2, num_filter=int(num_filter*0.25), kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, workspace=workspace, name=name + '_conv2', act_bit=ACT_BIT, weight_bit=bit) bn3 = mx.sym.BatchNorm(data=conv2, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn3') if not binarize: act3 = Act(data=bn3, act_type='relu', name=name + '_relu3') if not dcn: conv3 = Conv(data=act3, num_filter=int(num_filter*0.25), kernel=(3,3), stride=(1,1), pad=(dilation,dilation), dilate=(dilation,dilation), no_bias=True, workspace=workspace, name=name + '_conv3') else: conv3_offset = mx.symbol.Convolution(name=name+'_conv3_offset', data = act3, num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) conv3 = mx.contrib.symbol.DeformableConvolution(name=name+'_conv3', data=act3, offset=conv3_offset, num_filter=int(num_filter*0.25), pad=(1,1), kernel=(3, 3), num_deformable_group=1, stride=(1, 1), dilate=(1, 1), no_bias=True) else: act3 = mx.sym.QActivation(data=bn3, act_bit=ACT_BIT, name=name + '_relu3', backward_only=True) conv3 = mx.sym.QConvolution_v1(data=act3, num_filter=int(num_filter*0.25), kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, workspace=workspace, name=name + '_conv3', act_bit=ACT_BIT, weight_bit=bit) conv4 = mx.symbol.Concat(*[conv1, conv2, conv3]) if binarize: conv4 = mx.sym.BatchNorm(data=conv4, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_bn4') if dim_match: shortcut = data else: if not binarize: shortcut = Conv(data=act1, num_filter=num_filter, kernel=(1,1), stride=stride, no_bias=True, workspace=workspace, name=name+'_sc') else: #assert(False) shortcut = mx.sym.QConvolution_v1(data=act1, num_filter=num_filter, kernel=(1,1), stride=stride, pad=(0,0), no_bias=True, workspace=workspace, name=name + '_sc', act_bit=ACT_BIT, weight_bit=bit) shortcut = mx.sym.BatchNorm(data=shortcut, fix_gamma=False, eps=2e-5, momentum=bn_mom, name=name + '_sc_bn') if memonger: shortcut._set_attr(mirror_stage='True') return conv4 + shortcut #return bn4 + shortcut #return act4 + shortcut def block17(net, input_num_channels, scale=1.0, with_act=True, act_type='relu', mirror_attr={}, name=''): tower_conv = ConvFactory(net, 192, (1, 1), name=name+'_conv') tower_conv1_0 = ConvFactory(net, 129, (1, 1), name=name+'_conv1_0') tower_conv1_1 = ConvFactory(tower_conv1_0, 160, (1, 7), pad=(1, 2), name=name+'_conv1_1') tower_conv1_2 = ConvFactory(tower_conv1_1, 192, (7, 1), pad=(2, 1), name=name+'_conv1_2') tower_mixed = mx.symbol.Concat(*[tower_conv, tower_conv1_2]) tower_out = ConvFactory( tower_mixed, input_num_channels, (1, 1), with_act=False, name=name+'_conv_out') net = net+scale * tower_out if with_act: act = mx.symbol.Activation( data=net, act_type=act_type, attr=mirror_attr) return act else: return net def block35(net, input_num_channels, scale=1.0, with_act=True, act_type='relu', mirror_attr={}, name=''): M = 1.0 tower_conv = ConvFactory(net, int(input_num_channels*0.25*M), (1, 1), name=name+'_conv') tower_conv1_0 = ConvFactory(net, int(input_num_channels*0.25*M), (1, 1), name=name+'_conv1_0') tower_conv1_1 = ConvFactory(tower_conv1_0, int(input_num_channels*0.25*M), (3, 3), pad=(1, 1), name=name+'_conv1_1') tower_conv2_0 = ConvFactory(net, int(input_num_channels*0.25*M), (1, 1), name=name+'_conv2_0') tower_conv2_1 = ConvFactory(tower_conv2_0, int(input_num_channels*0.375*M), (3, 3), pad=(1, 1), name=name+'_conv2_1') tower_conv2_2 = ConvFactory(tower_conv2_1, int(input_num_channels*0.5*M), (3, 3), pad=(1, 1), name=name+'_conv2_2') tower_mixed = mx.symbol.Concat(*[tower_conv, tower_conv1_1, tower_conv2_2]) tower_out = ConvFactory( tower_mixed, input_num_channels, (1, 1), with_act=False, name=name+'_conv_out') net = net+scale * tower_out if with_act: act = mx.symbol.Activation( data=net, act_type=act_type, attr=mirror_attr) return act else: return net def conv_inception(data, num_filter, stride, dim_match, name, binarize, dcn, dilate, **kwargs): assert not binarize if stride[0]>1 or not dim_match: return conv_resnet(data, num_filter, stride, dim_match, name, binarize, dcn, dilate, **kwargs) conv4 = block35(data, num_filter, name=name+'_block35') return conv4 def conv_cab(data, num_filter, stride, dim_match, name, binarize, dcn, dilate, **kwargs): if stride[0]>1 or not dim_match: return conv_hpm(data, num_filter, stride, dim_match, name, binarize, dcn, dilate, **kwargs) cab = CAB(data, num_filter, 1, 4, workspace, name, dilate, 1) return cab.get() def conv_block(data, num_filter, stride, dim_match, name, binarize, dcn, dilate): if config.net_block=='resnet': return conv_resnet(data, num_filter, stride, dim_match, name, binarize, dcn, dilate) elif config.net_block=='inception': return conv_inception(data, num_filter, stride, dim_match, name, binarize, dcn, dilate) elif config.net_block=='hpm': return conv_hpm(data, num_filter, stride, dim_match, name, binarize, dcn, dilate) elif config.net_block=='cab': return conv_cab(data, num_filter, stride, dim_match, name, binarize, dcn, dilate) def hourglass(data, nFilters, nModules, n, workspace, name, binarize, dcn): s = 2 _dcn = False up1 = data for i in range(nModules): up1 = conv_block(up1, nFilters, (1,1), True, "%s_up1_%d"%(name,i), binarize, _dcn, 1) low1 = mx.sym.Pooling(data=data, kernel=(s, s), stride=(s,s), pad=(0,0), pool_type='max') for i in range(nModules): low1 = conv_block(low1, nFilters, (1,1), True, "%s_low1_%d"%(name,i), binarize, _dcn, 1) if n>1: low2 = hourglass(low1, nFilters, nModules, n-1, workspace, "%s_%d"%(name, n-1), binarize, dcn) else: low2 = low1 for i in range(nModules): low2 = conv_block(low2, nFilters, (1,1), True, "%s_low2_%d"%(name,i), binarize, _dcn, 1) #TODO low3 = low2 for i in range(nModules): low3 = conv_block(low3, nFilters, (1,1), True, "%s_low3_%d"%(name,i), binarize, _dcn, 1) up2 = mx.symbol.UpSampling(low3, scale=s, sample_type='nearest', workspace=512, name='%s_upsampling_%s'%(name,n), num_args=1) return mx.symbol.add_n(up1, up2) class STA: def __init__(self, data, nFilters, nModules, n, workspace, name): self.data = data self.nFilters = nFilters self.nModules = nModules self.n = n self.workspace = workspace self.name = name self.sym_map = {} def get_conv(self, data, name, dilate=1, group=1): cab = CAB(data, self.nFilters, self.nModules, 4, self.workspace, name, dilate, group) return cab.get() def get_output(self, w, h): #print(w,h) assert w>=1 and w<=config.net_n+1 assert h>=1 and h<=config.net_n+1 s = 2 bn_mom = 0.9 key = (w,h) if key in self.sym_map: return self.sym_map[key] ret = None if h==self.n: if w==self.n: ret = self.data,64 else: x = self.get_output(w+1, h) body = self.get_conv(x[0], "%s_w%d_h%d_1"%(self.name, w, h)) body = mx.sym.Pooling(data=body, kernel=(s, s), stride=(s,s), pad=(0,0), pool_type='max') body = self.get_conv(body, "%s_w%d_h%d_2"%(self.name, w, h)) ret = body, x[1]//2 else: x = self.get_output(w+1, h+1) y = self.get_output(w, h+1) HC = False if h%2==1 and h!=w: xbody = lin3(x[0], self.nFilters, self.workspace, "%s_w%d_h%d_x"%(self.name, w, h), 3, self.nFilters, 1) HC = True #xbody = x[0] else: xbody = x[0] if x[1]//y[1]==2: if w>1: ybody = mx.symbol.Deconvolution(data=y[0], num_filter=self.nFilters, kernel=(s,s), stride=(s, s), name='%s_upsampling_w%d_h%d'%(self.name,w, h), attr={'lr_mult': '1.0'}, workspace=self.workspace) ybody = mx.sym.BatchNorm(data=ybody, fix_gamma=False, momentum=bn_mom, eps=2e-5, name="%s_w%d_h%d_y_bn"%(self.name, w, h)) ybody = Act(data=ybody, act_type='relu', name="%s_w%d_h%d_y_act"%(self.name, w, h)) else: if h>=1: ybody = mx.symbol.UpSampling(y[0], scale=s, sample_type='nearest', workspace=512, name='%s_upsampling_w%d_h%d'%(self.name,w, h), num_args=1) ybody = self.get_conv(ybody, "%s_w%d_h%d_4"%(self.name, w, h)) else: ybody = mx.symbol.Deconvolution(data=y[0], num_filter=self.nFilters, kernel=(s,s), stride=(s, s), name='%s_upsampling_w%d_h%d'%(self.name,w, h), attr={'lr_mult': '1.0'}, workspace=self.workspace) ybody = mx.sym.BatchNorm(data=ybody, fix_gamma=False, momentum=bn_mom, eps=2e-5, name="%s_w%d_h%d_y_bn"%(self.name, w, h)) ybody = Act(data=ybody, act_type='relu', name="%s_w%d_h%d_y_act"%(self.name, w, h)) ybody = Conv(data=ybody, num_filter=self.nFilters, kernel=(3,3), stride=(1,1), pad=(1,1), no_bias=True, name="%s_w%d_h%d_y_conv2"%(self.name, w, h), workspace=self.workspace) ybody = mx.sym.BatchNorm(data=ybody, fix_gamma=False, momentum=bn_mom, eps=2e-5, name="%s_w%d_h%d_y_bn2"%(self.name, w, h)) ybody = Act(data=ybody, act_type='relu', name="%s_w%d_h%d_y_act2"%(self.name, w, h)) else: ybody = self.get_conv(y[0], "%s_w%d_h%d_5"%(self.name, w, h)) #if not HC: if config.net_sta==2 and h==3 and w==2: z = self.get_output(w+1, h) zbody = z[0] zbody = mx.sym.Pooling(data=zbody, kernel=(z[1], z[1]), stride=(z[1],z[1]), pad=(0,0), pool_type='avg') body = xbody+ybody body = body/2 body = mx.sym.broadcast_mul(body, zbody) else: #sta==1 body = xbody+ybody body = body/2 ret = body, x[1] assert ret is not None self.sym_map[key] = ret return ret def get(self): return self.get_output(1, 1)[0] class SymCoherent: def __init__(self, per_batch_size): self.per_batch_size = per_batch_size self.flip_order = [16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0, 26, 25, 24, 23, 22, 21, 20, 19, 18, 17, 27, 28, 29, 30, 35, 34, 33, 32, 31, 45, 44, 43, 42, 47, 46, 39, 38, 37, 36, 41, 40, 54, 53, 52, 51, 50, 49, 48, 59, 58, 57, 56, 55, 64, 63, 62, 61, 60, 67, 66, 65] def get(self, data): #data.shape[0]==per_batch_size b = self.per_batch_size//2 ux = mx.sym.slice_axis(data, axis=0, begin=0, end=b) dx = mx.sym.slice_axis(data, axis=0, begin=b, end=b*2) ux = mx.sym.flip(ux, axis=3) #ux = mx.sym.take(ux, indices = self.flip_order, axis=0) ux_list = [] for o in self.flip_order: _ux = mx.sym.slice_axis(ux, axis=1, begin=o, end=o+1) ux_list.append(_ux) ux = mx.sym.concat(*ux_list, dim=1) return ux, dx def l2_loss(x, y): loss = x-y loss = mx.symbol.smooth_l1(loss, scalar=1.0) #loss = loss*loss loss = mx.symbol.mean(loss) return loss def ce_loss(x, y): #loss = mx.sym.SoftmaxOutput(data = x, label = y, normalization='valid', multi_output=True) x_max = mx.sym.max(x, axis=[2,3], keepdims=True) x = mx.sym.broadcast_minus(x, x_max) body = mx.sym.exp(x) sums = mx.sym.sum(body, axis=[2,3], keepdims=True) body = mx.sym.broadcast_div(body, sums) loss = mx.sym.log(body) loss = loss*y*-1.0 loss = mx.symbol.mean(loss, axis=[1,2,3]) #loss = mx.symbol.mean(loss) return loss def get_symbol(num_classes): m = config.multiplier sFilters = max(int(64*m), 32) mFilters = max(int(128*m), 32) nFilters = int(256*m) nModules = 1 nStacks = config.net_stacks binarize = config.net_binarize input_size = config.input_img_size label_size = config.output_label_size use_coherent = config.net_coherent use_STA = config.net_sta N = config.net_n DCN = config.net_dcn per_batch_size = config.per_batch_size print('binarize', binarize) print('use_coherent', use_coherent) print('use_STA', use_STA) print('use_N', N) print('use_DCN', DCN) print('per_batch_size', per_batch_size) #assert(label_size==64 or label_size==32) #assert(input_size==128 or input_size==256) coherentor = SymCoherent(per_batch_size) D = input_size // label_size print(input_size, label_size, D) data = mx.sym.Variable(name='data') data = data-127.5 data = data*0.0078125 gt_label = mx.symbol.Variable(name='softmax_label') losses = [] closses = [] ref_label = gt_label if D==4: body = Conv(data=data, num_filter=sFilters, kernel=(7, 7), stride=(2,2), pad=(3, 3), no_bias=True, name="conv0", workspace=workspace) else: body = Conv(data=data, num_filter=sFilters, kernel=(3, 3), stride=(1,1), pad=(1, 1), no_bias=True, name="conv0", workspace=workspace) body = mx.sym.BatchNorm(data=body, fix_gamma=False, eps=2e-5, momentum=bn_mom, name='bn0') body = Act(data=body, act_type='relu', name='relu0') dcn = False body = conv_block(body, mFilters, (1,1), sFilters==mFilters, 'res0', False, dcn, 1) body = mx.sym.Pooling(data=body, kernel=(2, 2), stride=(2,2), pad=(0,0), pool_type='max') body = conv_block(body, mFilters, (1,1), True, 'res1', False, dcn, 1) #TODO body = conv_block(body, nFilters, (1,1), mFilters==nFilters, 'res2', binarize, dcn, 1) #binarize=True? heatmap = None for i in range(nStacks): shortcut = body if config.net_sta>0: sta = STA(body, nFilters, nModules, config.net_n+1, workspace, 'sta%d'%(i)) body = sta.get() else: body = hourglass(body, nFilters, nModules, config.net_n, workspace, 'stack%d_hg'%(i), binarize, dcn) for j in range(nModules): body = conv_block(body, nFilters, (1,1), True, 'stack%d_unit%d'%(i,j), binarize, dcn, 1) _dcn = True if config.net_dcn>=2 else False ll = ConvFactory(body, nFilters, (1,1), dcn = _dcn, name='stack%d_ll'%(i)) _name = "heatmap%d"%(i) if i=2 else False if not _dcn: out = Conv(data=ll, num_filter=num_classes, kernel=(1, 1), stride=(1,1), pad=(0,0), name=_name, workspace=workspace) else: out_offset = mx.symbol.Convolution(name=_name+'_offset', data = ll, num_filter=18, pad=(1, 1), kernel=(3, 3), stride=(1, 1)) out = mx.contrib.symbol.DeformableConvolution(name=_name, data=ll, offset=out_offset, num_filter=num_classes, pad=(1,1), kernel=(3, 3), num_deformable_group=1, stride=(1, 1), dilate=(1, 1), no_bias=False) #out = Conv(data=ll, num_filter=num_classes, kernel=(3,3), stride=(1,1), pad=(1,1), # name=_name, workspace=workspace) if i==nStacks-1: heatmap = out loss = ce_loss(out, ref_label) #loss = loss/nStacks #loss = l2_loss(out, ref_label) losses.append(loss) if config.net_coherent>0: ux, dx = coherentor.get(out) closs = l2_loss(ux, dx) closs = closs/nStacks closses.append(closs) if i0: coherent_weight = 0.0001 closs = mx.symbol.add_n(*closses) closs = mx.symbol.MakeLoss(closs, grad_scale = coherent_weight) syms.append(closs) syms.append(pred) sym = mx.symbol.Group( syms ) return sym def init_weights(sym, data_shape_dict): #print('in hg') arg_name = sym.list_arguments() aux_name = sym.list_auxiliary_states() arg_shape, _, aux_shape = sym.infer_shape(**data_shape_dict) arg_shape_dict = dict(zip(arg_name, arg_shape)) aux_shape_dict = dict(zip(aux_name, aux_shape)) #print(aux_shape) #print(aux_params) #print(arg_shape_dict) arg_params = {} aux_params = {} for k in arg_shape_dict: v = arg_shape_dict[k] #print(k,v) if k.endswith('offset_weight') or k.endswith('offset_bias'): print('initializing',k) arg_params[k] = mx.nd.zeros(shape = v) elif k.startswith('fc6_'): if k.endswith('_weight'): print('initializing',k) arg_params[k] = mx.random.normal(0, 0.01, shape=v) elif k.endswith('_bias'): print('initializing',k) arg_params[k] = mx.nd.zeros(shape=v) elif k.find('upsampling')>=0: print('initializing upsampling_weight', k) arg_params[k] = mx.nd.zeros(shape=arg_shape_dict[k]) init = mx.init.Initializer() init._init_bilinear(k, arg_params[k]) return arg_params, aux_params