print(
param[im]
) # parameter in form tensor([2.5508, 0.0000, 0.0000, 0.0000, 0.0000, 5.0000])
image2show = image[im] # indexing random one image
print(image2show.size()) #torch.Size([3, 512, 512])
plt.imshow((image2show * 0.5 + 0.5).numpy().transpose(1, 2, 0))
plt.show()
break # break here just to show 1 batch of data
# ------------------------------------------------------------------
# for noise in np.arange(0, 1, 0.1):
noise = 0.0
model = resnet50(cifar=True) #use pretrained on imagenet if cifar is true
model = model.to(device) # transfer the neural net onto the GPU
criterion = nn.MSELoss(
) # define the loss (MSE, Crossentropy, Binarycrossentropy)
# ------------------------------------------------------------------
all_Train_losses, all_Test_losses = train(model, train_dataloader,
test_dataloader, n_epochs, criterion,
date4File, cubeSetName, batch_size,
fileExtension, device, noise)
# ------------------------------------------------------------------
torch.save(
model.state_dict(),
#
# # print(image[2])
# print(image.size(), param.size()) #torch.Size([batch, 3, 512, 512]) torch.Size([batch, 6])
# im =2
# print(param[im]) # parameter in form tensor([2.5508, 0.0000, 0.0000, 0.0000, 0.0000, 5.0000])
#
# image2show = image[im] # indexing random one image
# print(image2show.size()) #torch.Size([3, 512, 512])
# plt.imshow((image2show * 0.5 + 0.5).numpy().transpose(1, 2, 0))
# plt.show()
# break # break here just to show 1 batch of data
# ------------------------------------------------------------------
model = resnet50(
cifar=False,
modelName=modelName) #train with the saved model from the training script
model = model.to(device) # transfer the neural net onto the GPU
criterion = nn.MSELoss()
# ------------------------------------------------------------------
# test the model
parameters, predicted_params, test_losses, al, bl, gl, xl, yl, zl = testRenderResnet(
model, test_dataloader, criterion, file_name_extension, device, obj_name)
# ------------------------------------------------------------------
# display computed parameter against ground truth
obj_name = 'rubik_color'
# plt.show()
#
# image2show = sil[im] # indexing random one image
# print(image2show.size()) # torch.Size([3, 512, 512])
# image2show = image2show.numpy()
# plt.imshow(image2show, cmap='gray')
# plt.show()
#
# break # break here just to show 1 batch of data
# ------------------------------------------------------------------
# for noise in np.arange(0, 1, 0.1):
noise = 0.0
# model = resnet50(cifar=False, modelName=modelName) #train with the saved model from the training script
model = resnet50(cifar=True) #train with the pretrained parameter from cifar database
# model = resnet50_multCPU(cifar=True)
model = model.to(device) # transfer the neural net onto the GPU
criterion = nn.MSELoss() #nn.BCELoss() #nn.CrossEntropyLoss() define the loss (MSE, Crossentropy, Binarycrossentropy)
#
# ------------------------------------------------------------------
train_losses, all_Test_losses = train_render(model, train_dataloader, test_dataloader,
n_epochs, criterion,
date4File, cubeSetName, batch_size, fileExtension, device, obj_name, noise)
# ------------------------------------------------------------------
torch.save(model.state_dict(), 'models/{}_FinalModel_train_{}_{}batchs_{}epochs_Noise{}_{}_RenderRegr.pth'.format(date4File, cubeSetName, str(batch_size), str(n_epochs), noise*100,fileExtension))
print('parameters saved')