train_H3D.py

#
#
#      0=================================0
#      |    Kernel Point Convolutions    |
#      0=================================0
#
#
# ----------------------------------------------------------------------------------------------------------------------
#
#      Callable script to start a training on H3D dataset
#
# ----------------------------------------------------------------------------------------------------------------------

# ----------------------------------------------------------------------------------------------------------------------
#
#           Imports and global variables
#       \**********************************/
#

# Common libs
import signal

# Dataset
from datasets.H3D import *
from torch.utils.data import DataLoader

from utils.config import Config
from utils.trainer import ModelTrainer
from models.architectures import KPFCNN

# ----------------------------------------------------------------------------------------------------------------------
#
#           Config Class
#       \******************/
#

class H3DConfig(Config):
    """
    Override the parameters you want to modify for this dataset
    """

    ####################
    # Dataset parameters
    ####################

    # Dataset name
    dataset = 'H3D'

    # Number of classes in the dataset (This value is overwritten by dataset class when Initializating dataset).
    num_classes = None

    # Type of task performed on this dataset (also overwritten)
    dataset_task = ''

    # Number of CPU threads for the input pipeline
    input_threads = 12

    #########################
    # Architecture definition
    #########################

    # Define layers
    architecture = ['simple',
                    'resnetb',
                    'resnetb_strided',
                    'resnetb',
                    'resnetb_strided',
                    'resnetb',
                    'resnetb_strided',
                    'resnetb',
                    'resnetb_strided',
                    'resnetb',
                    'nearest_upsample',
                    'unary',
                    'nearest_upsample',
                    'unary',
                    'nearest_upsample',
                    'unary',
                    'nearest_upsample',
                    'unary'
                    ]

    ###################
    # KPConv parameters
    ###################

    # Radius of the input sphere
    in_radius = 5.0

    # Number of kernel points
    num_kernel_points = 15

    # Size of the first subsampling grid in meter
    first_subsampling_dl = 0.1

    # Radius of convolution in "number grid cell". (2.5 is the standard value)
    conv_radius = 2.5

    # Radius of deformable convolution in "number grid cell". Larger so that deformed kernel can spread out
    deform_radius = 5.0

    # Radius of the area of influence of each kernel point in "number grid cell". (1.0 is the standard value)
    KP_extent = 1.0

    # Behavior of convolutions in ('constant', 'linear', 'gaussian')
    KP_influence = 'linear'

    # Aggregation function of KPConv in ('closest', 'sum')
    aggregation_mode = 'sum'

    # Choice of input features
    first_features_dim = 64
    in_features_dim = 4

    # Can the network learn modulations
    modulated = False

    # Batch normalization parameters
    use_batch_norm = True
    batch_norm_momentum = 0.02

    # Deformable offset loss
    # 'point2point' fitting geometry by penalizing distance from deform point to input points
    # 'point2plane' fitting geometry by penalizing distance from deform point to input point triplet (not implemented)
    deform_fitting_mode = 'point2point'
    deform_fitting_power = 1.0              # Multiplier for the fitting/repulsive loss
    deform_lr_factor = 0.1                  # Multiplier for learning rate applied to the deformations
    repulse_extent = 1.2                    # Distance of repulsion for deformed kernel points

    #####################
    # Training parameters
    #####################

    # Maximal number of epochs
    max_epoch = 200

    # Learning rate management
    learning_rate = 1e-3
    momentum = 0.98
    lr_decays = {i: 0.1 ** (1 / 100) for i in range(1, max_epoch+1)}
    grad_clip_norm = 100.0

    # Number of batch
    batch_num = 1
    batch_limit = 150000

    # Number of steps per epochs
    epoch_steps = 400

    # Number of validation examples per epoch
    validation_size = 20

    # Number of epoch between each checkpoint
    checkpoint_gap = 20

    # Augmentations
    augment_scale_anisotropic = True
    augment_symmetries = [True, False, False]
    augment_rotation = 'vertical'
    augment_scale_min = 0.9
    augment_scale_max = 1.1
    augment_noise = 0.02
    augment_color = 1.0

    # The way we balance segmentation loss
    #   > 'none': Each point in the whole batch has the same contribution.
    #   > 'class': Each class has the same contribution (points are weighted according to class balance)
    #   > 'batch': Each cloud in the batch has the same contribution (points are weighted according cloud sizes)
    segloss_balance = 'class'

    # Do we nee to save convergence
    saving = True
    saving_path = None

    weak_level = 1 # number of labeled points for each class in sub-clouds, 1 for one class one click
    al_itr = 5 # number of iterations for active learning
    al_initnum = 150 # number of initial samples
    al_num = 150 # number of samples to be added each time
    acc_thr = 0.98 # accuracy threshold for early stopping
    serial = '1' # serial number of the experiment for robustness test

    test_mode = False


# ----------------------------------------------------------------------------------------------------------------------
#
#           Main Call
#       \***************/
#

if __name__ == '__main__':

    ############################
    # Initialize the environment
    ############################

    # Set which gpu is going to be used
    GPU_ID = '0'

    # Set GPU visible device
    os.environ['CUDA_VISIBLE_DEVICES'] = GPU_ID

    ###############
    # Previous chkp
    ###############

    # Initialize configuration class
    config = H3DConfig()

    # Choose here if you want to start training from a previous snapshot (None for new training)
    config.previous_training_path = ''
    if config.previous_training_path:
        # config.load(os.path.join('results', config.dataset, config.previous_training_path))
        config.saving_path = None

    # Choose index of checkpoint to start from. If None, uses the latest chkp
    config.chkp_idx = -1
    if config.previous_training_path:

        # Find all snapshot in the chosen training folder
        chkp_path = os.path.join('results', config.dataset, config.previous_training_path, 'checkpoints')
        chkps = [f for f in os.listdir(chkp_path) if f[:4] == 'chkp']

        # Find which snapshot to restore
        if config.chkp_idx is None:
            chosen_chkp = 'current_chkp.tar'
        else:
            chosen_chkp = np.sort(chkps)[config.chkp_idx]
        chosen_chkp = os.path.join('results', config.dataset, config.previous_training_path, 'checkpoints', chosen_chkp)

    else:
        chosen_chkp = None

    ##############
    # Prepare Data
    ##############

    print()
    print('Data Preparation')
    print('****************')


    # Initialize datasets
    training_dataset = H3DDataset(config, set='training', use_potentials=True)
    test_dataset = H3DDataset(config, set='validation', use_potentials=True)

    # Initialize samplers
    training_sampler = H3DSampler(training_dataset)
    test_sampler = H3DSampler(test_dataset)

    # Initialize the dataloader
    training_loader = DataLoader(training_dataset,
                                 batch_size=1,
                                 sampler=training_sampler,
                                 collate_fn=H3DCollate,
                                 num_workers=config.input_threads,
                                 pin_memory=True)
    test_loader = DataLoader(test_dataset,
                             batch_size=1,
                             sampler=test_sampler,
                             collate_fn=H3DCollate,
                             num_workers=config.input_threads,
                             pin_memory=True)

    # Set to use all subclouds for calibration
    temp = training_dataset.config.weak_level
    training_dataset.config.weak_level = None
    # Calibrate samplers
    training_sampler.calibration(training_loader, verbose=True)
    # Reset to normal mode
    training_dataset.config.weak_level = temp


    print('\nModel Preparation')
    print('*****************')

    # Define network model
    t1 = time.time()
    net = KPFCNN(config, training_dataset.label_values, training_dataset.ignored_labels)

    debug = False
    if debug:
        # print('\n*************************************\n')
        # print(net)
        # print('\n*************************************\n')
        # for param in net.parameters():
        #     if param.requires_grad:
                # print(param.shape)
        print('\n*************************************\n')
        print("Model size %i" % sum(param.numel() for param in net.parameters() if param.requires_grad))
        print('\n*************************************\n')

    # Define a trainer class
    trainer = ModelTrainer(net, config, chkp_path=chosen_chkp)
    print('Done in {:.1f}s\n'.format(time.time() - t1))

    print('\nStart training')
    print('**************')

    # Training
    trainer.train(net, training_loader, test_loader, config)

    print('Forcing exit now')
    os.kill(os.getpid(), signal.SIGINT)