In file binary segmentation (camvid).ipynb, block 5, there is:

# Lets look at data we have
dataset = Dataset(x_train_dir, y_train_dir, classes=['car', 'pedestrian'])
image, mask = dataset[5] # get some sample
visualize(
    image=image, 
    cars_mask=mask[..., 0].squeeze(),
    sky_mask=mask[..., 1].squeeze(),
    background_mask=mask[..., 2].squeeze(),
)

here, sky_mask

Description
In some rare cases, for example, when you need to finetune a large model on a small dataset the majoring part of training loop is waiting for saving model checkpoints to a hard drive.

Proposal
Would be logically to add a CheckpointCallback with parameter save_n_best=0 to a configuration and do not store best checkpoints and instead use the latest state of the model.

Keras-rcnn was written to be compatible with a number of third-party frameworks and services like Apple’s Core ML framework that enables developers to embed Keras models into their iOS applications. We should document how an Apple developer can create, train, and export their model to their Core ML-compatible iOS application.

如果设置
cfg.NUM_TRAINERS = 4
cfg.TRAINER_ID = 0，1，2，3
if self.shuffle and cfg.NUM_TRAINERS > 1: np.random.RandomState(self.shuffle_seed).shuffle(self.all_lines) num_lines = len(self.all_lines) // cfg.NUM_TRAINERS self.lines = self.all_lines[num_lines * cfg.TRAINER_ID: num_lines * (cfg.TRAINER_ID + 1)] self.shuffle_seed += 1
上面代码中的self.

during CI on my hiwenet, I discovered a ton of deprecation warnings from scipy : https://travis-ci.org/github/raamana/hiwenet/jobs/658810855

mostly asking medpy to move from scipy to numpy for functions such as sum, absolute, sqrt, square etc .. see below for a sample, and the above link for a full list.. If you can, can you try fix them? thanks.

  /hom