274 lines
9.3 KiB
Python
274 lines
9.3 KiB
Python
|
|
||
|
import torch
|
||
|
import torch.nn as nn
|
||
|
import torch.nn.functional as F
|
||
|
import time
|
||
|
import timm
|
||
|
def filter(opState, kernelsize=5):
|
||
|
#B*100*2
|
||
|
|
||
|
Bs = opState.shape[0]
|
||
|
ches = opState.shape[1]
|
||
|
recL = int((kernelsize-3)/2)
|
||
|
labelTable = torch.zeros(Bs, int(ches+2*recL),opState.shape[2]).cuda()
|
||
|
labelTable[:,:recL,:] = opState[:,0,:].unsqueeze(dim=1)
|
||
|
labelTable[:,-recL:,:] = opState[:,-1,:].unsqueeze(dim=1)
|
||
|
labelTable[:,recL:-recL,:] = opState
|
||
|
|
||
|
newOpState = torch.zeros_like(opState)
|
||
|
|
||
|
tmpT = labelTable.unfold(1, kernelsize, 1)
|
||
|
tmpMeanT = torch.mean(tmpT, dim=-1)
|
||
|
newOpState[:,1:-1,:] = tmpMeanT
|
||
|
|
||
|
|
||
|
newOpState[:,0,:] = opState[:,0,:]
|
||
|
newOpState[:,-1,:] = opState[:,-1,:]
|
||
|
|
||
|
|
||
|
return newOpState
|
||
|
class BasicBlock(nn.Module):
|
||
|
expansion = 1
|
||
|
|
||
|
def __init__(self, in_planes, planes, stride=1):
|
||
|
super(BasicBlock, self).__init__()
|
||
|
self.conv1 = nn.Conv2d(
|
||
|
in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
|
||
|
self.bn1 = nn.BatchNorm2d(planes)
|
||
|
self.conv2 = nn.Conv2d(planes, planes, kernel_size=3,
|
||
|
stride=1, padding=1, bias=False)
|
||
|
self.bn2 = nn.BatchNorm2d(planes)
|
||
|
|
||
|
self.shortcut = nn.Sequential()
|
||
|
if stride != 1 or in_planes != self.expansion*planes:
|
||
|
self.shortcut = nn.Sequential(
|
||
|
nn.Conv2d(in_planes, self.expansion*planes,
|
||
|
kernel_size=1, stride=stride, bias=False),
|
||
|
nn.BatchNorm2d(self.expansion*planes)
|
||
|
)
|
||
|
|
||
|
def forward(self, x):
|
||
|
out = F.relu(self.bn1(self.conv1(x)))
|
||
|
out = self.bn2(self.conv2(out))
|
||
|
out += self.shortcut(x)
|
||
|
out = F.relu(out)
|
||
|
return out
|
||
|
|
||
|
|
||
|
class Bottleneck(nn.Module):
|
||
|
expansion = 4
|
||
|
|
||
|
def __init__(self, in_planes, planes, stride=1):
|
||
|
super(Bottleneck, self).__init__()
|
||
|
self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=1, bias=False)
|
||
|
self.bn1 = nn.BatchNorm2d(planes)
|
||
|
self.conv2 = nn.Conv2d(planes, planes, kernel_size=3,
|
||
|
stride=stride, padding=1, bias=False)
|
||
|
self.bn2 = nn.BatchNorm2d(planes)
|
||
|
self.conv3 = nn.Conv2d(planes, self.expansion *
|
||
|
planes, kernel_size=1, bias=False)
|
||
|
self.bn3 = nn.BatchNorm2d(self.expansion*planes)
|
||
|
|
||
|
self.shortcut = nn.Sequential()
|
||
|
if stride != 1 or in_planes != self.expansion*planes:
|
||
|
self.shortcut = nn.Sequential(
|
||
|
nn.Conv2d(in_planes, self.expansion*planes,
|
||
|
kernel_size=1, stride=stride, bias=False),
|
||
|
nn.BatchNorm2d(self.expansion*planes)
|
||
|
)
|
||
|
|
||
|
def forward(self, x):
|
||
|
out = F.relu(self.bn1(self.conv1(x)))
|
||
|
out = F.relu(self.bn2(self.conv2(out)))
|
||
|
out = self.bn3(self.conv3(out))
|
||
|
out += self.shortcut(x)
|
||
|
out = F.relu(out)
|
||
|
return out
|
||
|
|
||
|
|
||
|
class ResNet(nn.Module):
|
||
|
def __init__(self, block, num_blocks, num_classes=600):
|
||
|
super(ResNet, self).__init__()
|
||
|
self.in_planes = 64
|
||
|
|
||
|
self.conv1 = nn.Conv2d(4, 64, kernel_size=3,
|
||
|
stride=1, padding=1, bias=False)
|
||
|
self.bn1 = nn.BatchNorm2d(64)
|
||
|
self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1)
|
||
|
self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2)
|
||
|
self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2)
|
||
|
self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2)
|
||
|
self.linear = nn.Linear(512*block.expansion, num_classes)
|
||
|
|
||
|
def _make_layer(self, block, planes, num_blocks, stride):
|
||
|
strides = [stride] + [1]*(num_blocks-1)
|
||
|
layers = []
|
||
|
for stride in strides:
|
||
|
layers.append(block(self.in_planes, planes, stride))
|
||
|
self.in_planes = planes * block.expansion
|
||
|
return nn.Sequential(*layers)
|
||
|
|
||
|
def forward(self, x):
|
||
|
out = F.relu(self.bn1(self.conv1(x)))
|
||
|
out = self.layer1(out)
|
||
|
out = self.layer2(out)
|
||
|
out = self.layer3(out)
|
||
|
out = self.layer4(out)
|
||
|
out = F.avg_pool2d(out, 4)
|
||
|
out = out.view(out.size(0), -1)
|
||
|
out = self.linear(out)
|
||
|
return out
|
||
|
|
||
|
|
||
|
def ResNet18():
|
||
|
return ResNet(BasicBlock, [2, 2, 2, 2])
|
||
|
|
||
|
|
||
|
def ResNet34():
|
||
|
return ResNet(BasicBlock, [3, 4, 6, 3])
|
||
|
|
||
|
|
||
|
def ResNet50():
|
||
|
return ResNet(Bottleneck, [3, 4, 6, 3])
|
||
|
|
||
|
|
||
|
def ResNet101():
|
||
|
return ResNet(Bottleneck, [3, 4, 23, 3])
|
||
|
|
||
|
|
||
|
def ResNet152():
|
||
|
return ResNet(Bottleneck, [3, 8, 36, 3])
|
||
|
|
||
|
|
||
|
def test():
|
||
|
net = ResNet18()
|
||
|
y = net(torch.randn(1, 3, 32, 32))
|
||
|
print(y.size())
|
||
|
|
||
|
|
||
|
class resnet101(nn.Module):
|
||
|
def __init__(self, filter = 7):
|
||
|
super(resnet101, self).__init__()
|
||
|
self.image = timm.create_model('resnet101',num_classes=600)
|
||
|
self.image._modules['conv1'] = nn.Conv2d(4, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
|
||
|
# print(self.image)
|
||
|
self.filter = filter
|
||
|
|
||
|
def forward(self, x, labelState, labelRot):
|
||
|
Bs = x.shape[0]
|
||
|
ft = torch.reshape(self.image(x),(Bs,200,3))
|
||
|
# print("lo.shape", lo.shape)
|
||
|
opState = torch.zeros_like(labelState)
|
||
|
yw = ft[:,:,2]
|
||
|
cosyaw = torch.cos(yw).unsqueeze(dim=2) #b*100*1
|
||
|
sinyaw = torch.sin(yw).unsqueeze(dim=2) #b*100*1
|
||
|
rotOutput = torch.cat((cosyaw,sinyaw), dim=2) #b*100*2
|
||
|
opState[:,1:-1,:] = ft[:,1:-1,0:2]
|
||
|
opState[:,0,:] = labelState[:,0,:]
|
||
|
opState[:,-1,:] = labelState[:,-1,:]
|
||
|
# print("rotOutput.shape", rotOutput.shape)
|
||
|
# print("labelRot.shape", labelRot.shape)
|
||
|
rotOutput[:,0,:] = labelRot[:,0,:]
|
||
|
rotOutput[:,-1,:] = labelRot[:,-1,:]
|
||
|
if(self.filter >=3):
|
||
|
opState = filter(opState, self.filter)
|
||
|
rotOutput = filter(rotOutput, self.filter)#B*99*2
|
||
|
rotOutput = torch.nn.functional.normalize(rotOutput, dim=2)
|
||
|
|
||
|
|
||
|
return opState, rotOutput
|
||
|
|
||
|
class resnet50(nn.Module):
|
||
|
def __init__(self, filter = 7):
|
||
|
super(resnet50, self).__init__()
|
||
|
self.image = timm.create_model('resnet50',num_classes=600)
|
||
|
self.image._modules['conv1'] = nn.Conv2d(4, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
|
||
|
# print(self.image)
|
||
|
self.filter = filter
|
||
|
|
||
|
def forward(self, x, labelState, labelRot):
|
||
|
Bs = x.shape[0]
|
||
|
ft = torch.reshape(self.image(x),(Bs,200,3))
|
||
|
# print("lo.shape", lo.shape)
|
||
|
opState = torch.zeros_like(labelState)
|
||
|
yw = ft[:,:,2]
|
||
|
cosyaw = torch.cos(yw).unsqueeze(dim=2) #b*100*1
|
||
|
sinyaw = torch.sin(yw).unsqueeze(dim=2) #b*100*1
|
||
|
rotOutput = torch.cat((cosyaw,sinyaw), dim=2) #b*100*2
|
||
|
opState[:,1:-1,:] = ft[:,1:-1,0:2]
|
||
|
opState[:,0,:] = labelState[:,0,:]
|
||
|
opState[:,-1,:] = labelState[:,-1,:]
|
||
|
# print("rotOutput.shape", rotOutput.shape)
|
||
|
# print("labelRot.shape", labelRot.shape)
|
||
|
rotOutput[:,0,:] = labelRot[:,0,:]
|
||
|
rotOutput[:,-1,:] = labelRot[:,-1,:]
|
||
|
if(self.filter >=3):
|
||
|
opState = filter(opState, self.filter)
|
||
|
rotOutput = filter(rotOutput, self.filter)#B*99*2
|
||
|
rotOutput = torch.nn.functional.normalize(rotOutput, dim=2)
|
||
|
|
||
|
|
||
|
return opState, rotOutput
|
||
|
|
||
|
class resnet152(nn.Module):
|
||
|
def __init__(self, filter = 7):
|
||
|
super(resnet152, self).__init__()
|
||
|
self.image = timm.create_model('resnet152',num_classes=600)
|
||
|
self.image._modules['conv1'] = nn.Conv2d(4, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
|
||
|
# print(self.image)
|
||
|
self.filter = filter
|
||
|
|
||
|
def forward(self, x, labelState, labelRot):
|
||
|
Bs = x.shape[0]
|
||
|
ft = torch.reshape(self.image(x),(Bs,200,3))
|
||
|
# print("lo.shape", lo.shape)
|
||
|
opState = torch.zeros_like(labelState)
|
||
|
yw = ft[:,:,2]
|
||
|
cosyaw = torch.cos(yw).unsqueeze(dim=2) #b*100*1
|
||
|
sinyaw = torch.sin(yw).unsqueeze(dim=2) #b*100*1
|
||
|
rotOutput = torch.cat((cosyaw,sinyaw), dim=2) #b*100*2
|
||
|
opState[:,1:-1,:] = ft[:,1:-1,0:2]
|
||
|
opState[:,0,:] = labelState[:,0,:]
|
||
|
opState[:,-1,:] = labelState[:,-1,:]
|
||
|
# print("rotOutput.shape", rotOutput.shape)
|
||
|
# print("labelRot.shape", labelRot.shape)
|
||
|
rotOutput[:,0,:] = labelRot[:,0,:]
|
||
|
rotOutput[:,-1,:] = labelRot[:,-1,:]
|
||
|
if(self.filter >=3):
|
||
|
opState = filter(opState, self.filter)
|
||
|
rotOutput = filter(rotOutput, self.filter)#B*99*2
|
||
|
rotOutput = torch.nn.functional.normalize(rotOutput, dim=2)
|
||
|
|
||
|
|
||
|
return opState, rotOutput
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
|
||
|
# test()
|
||
|
if __name__ == "__main__":
|
||
|
# test()
|
||
|
pt = 200
|
||
|
model = resnet152(pt_num=pt, filter=7).cuda().half()
|
||
|
totalt = 0.0
|
||
|
count = 0
|
||
|
model.eval()
|
||
|
# out = model(input)
|
||
|
input = torch.rand(1,4,200,200).cuda().half()
|
||
|
labelState = torch.rand(1,pt,2).cuda().half()
|
||
|
labelRot = torch.rand(1,pt,2).cuda().half()
|
||
|
anchors = torch.rand(1,pt,20,20).cuda().half()
|
||
|
with torch.no_grad():
|
||
|
for i in range(100):
|
||
|
torch.cuda.synchronize()
|
||
|
start = time.time()
|
||
|
out = model(input, labelState, labelRot, anchors)
|
||
|
torch.cuda.synchronize()
|
||
|
end = time.time()
|
||
|
if i>=10:
|
||
|
totalt += 1000.0*(end-start)
|
||
|
count +=1
|
||
|
print("model time: ", totalt / count, " ms")
|
||
|
|
||
|
|