from processors import SimpleDataGenerator
from readers import KittiDataReader
tf.get_logger().setLevel("ERROR")
DATA_ROOT = "../training" # TODO make main arg
MODEL_ROOT = "./logs"
os.environ["CUDA_DEVICE_ORDER"] = "PCI_BUS_ID"
if __name__ == "__main__":
params = Parameters()
gpus = tf.config.experimental.list_physical_devices('GPU')
loss = PointPillarNetworkLoss(params)
optimizer = tf.keras.optimizers.Adam(lr=params.learning_rate,
decay=params.decay_rate)
if len(gpus) > 1:
strategy = tf.distribute.MirroredStrategy(
cross_device_ops=tf.distribute.HierarchicalCopyAllReduce())
with strategy.scope():
pillar_net = build_point_pillar_graph(params)
pillar_net.load_weights(os.path.join(MODEL_ROOT, "model.h5"))
pillar_net.compile(optimizer, loss=loss.losses())
else:
pillar_net = build_point_pillar_graph(params)
pillar_net.load_weights(os.path.join(MODEL_ROOT, "model.h5"))
pillar_net.compile(optimizer, loss=loss.losses())
data_reader = KittiDataReader()
from loss import PointPillarNetworkLoss
from network import build_point_pillar_graph
from processors import SimpleDataGenerator
from readers import KittiDataReader
DATA_ROOT = "/Users/chirag/Documents/Projects/682/PointPillars/data" # TODO make main arg
TEST_ROOT = "/Users/chirag/Documents/Projects/682/PointPillars/data/testing" # TODO make main arg
if __name__ == "__main__":
params = Parameters()
pillar_net = build_point_pillar_graph(params)
loss = PointPillarNetworkLoss(params)
optimizer = tf.keras.optimizers.Adam(lr=params.learning_rate, decay=params.decay_rate)
pillar_net.compile(optimizer, loss=loss.losses(),metrics=['accuracy'])
log_dir = "./logs"
callbacks = [
tf.keras.callbacks.TensorBoard(log_dir=log_dir),
tf.keras.callbacks.ModelCheckpoint(filepath=os.path.join(log_dir, "model.h5"), save_best_only=True),
tf.keras.callbacks.LearningRateScheduler(
lambda epoch, lr: lr * 0.8 if ((epoch % 15 == 0) and (epoch != 0)) else lr, verbose=True),
tf.keras.callbacks.EarlyStopping(patience=20),
]
data_reader = KittiDataReader()