基本信息

---

下载

   友情提示：（无需注册或充值，赞助后即可获取资源下载链接）

---

     嘿，亲！知识可是无价之宝呢，但咱这精心整理的资料也耗费了不少心血呀。小小地破费一下，绝对物超所值哦！如有下载和支付问题，请联系我们QQ(微信同号)：813200300

本次赞助数额为： 2 元　

支付宝支付

微信支付

   源码介绍

---

基于深度学习开发的自主避障算法

def trainModel(train_data_generator, val_data_generator, model, initial_epoch):
    """
    Model training.

    # Arguments
       train_data_generator: Training data generated batch by batch.
       val_data_generator: Validation data generated batch by batch.
       model: Target image channels.
       initial_epoch: Dimension of model output.
    """

    # Initialize loss weights
    model.alpha = tf.Variable(1, trainable=False, name='alpha', dtype=tf.float32)
    model.beta = tf.Variable(0, trainable=False, name='beta', dtype=tf.float32)

    # Initialize number of samples for hard-mining
    model.k_mse = tf.Variable(FLAGS.batch_size, trainable=False, name='k_mse', dtype=tf.int32)
    model.k_entropy = tf.Variable(FLAGS.batch_size, trainable=False, name='k_entropy', dtype=tf.int32)


    optimizer = optimizers.Adam(decay=1e-5)

    # Configure training process
    model.compile(loss=[utils.hard_mining_mse(model.k_mse),
                        utils.hard_mining_entropy(model.k_entropy)],
                        optimizer=optimizer, loss_weights=[model.alpha, model.beta])

    # Save model with the lowest validation loss
    weights_path = os.path.join(FLAGS.experiment_rootdir, 'weights_{epoch:03d}.h5')
    writeBestModel = ModelCheckpoint(filepath=weights_path, monitor='val_loss',
                                     save_best_only=True, save_weights_only=True)

    # Save model every 'log_rate' epochs.
    # Save training and validation losses.
    logz.configure_output_dir(FLAGS.experiment_rootdir)
    saveModelAndLoss = log_utils.MyCallback(filepath=FLAGS.experiment_rootdir,
                                            period=FLAGS.log_rate,
                                            batch_size=FLAGS.batch_size)

    # Train model
    steps_per_epoch = int(np.ceil(train_data_generator.samples / FLAGS.batch_size))
    validation_steps = int(np.ceil(val_data_generator.samples / FLAGS.batch_size))

    model.fit_generator(train_data_generator,
                        epochs=FLAGS.epochs, steps_per_epoch = steps_per_epoch,
                        callbacks=[writeBestModel, saveModelAndLoss],
                        validation_data=val_data_generator,
                        validation_steps = validation_steps,
                        initial_epoch=initial_epoch)