def save_predictions(train_ids, train_predictions, meta_test,
out_of_fold_test_predictions):
averaged_mask_predictions_test = np.mean(
np.array(out_of_fold_test_predictions), axis=0)
pipeline_postprocessing = pipelines.mask_postprocessing(config=CONFIG)
pipeline_postprocessing.clean_cache()
test_pipe_masks = {
'input_masks': {
'resized_images': averaged_mask_predictions_test
}
}
y_pred_test = pipeline_postprocessing.transform(
test_pipe_masks)['binarized_images']
LOGGER.info('Saving predictions')
out_of_fold_train_predictions_path = os.path.join(
EXPERIMENT_DIR, 'out_of_fold_train_predictions.pkl')
joblib.dump({
'ids': train_ids,
'images': train_predictions
}, out_of_fold_train_predictions_path)
out_of_fold_test_predictions_path = os.path.join(
EXPERIMENT_DIR, 'out_of_fold_test_predictions.pkl')
joblib.dump(
{
'ids': meta_test[ID_COLUMN].tolist(),
'images': averaged_mask_predictions_test
}, out_of_fold_test_predictions_path)
submission = utils.create_submission(meta_test, y_pred_test)
submission_filepath = os.path.join(EXPERIMENT_DIR, 'submission.csv')
submission.to_csv(submission_filepath, index=None, encoding='utf-8')
LOGGER.info('submission saved to {}'.format(submission_filepath))
LOGGER.info('submission head \n\n{}'.format(submission.head()))
def predict():
meta = pd.read_csv(PARAMS.metadata_filepath)
meta_test = meta[meta['is_train'] == 0]
if DEV_MODE:
meta_test = meta_test.sample(PARAMS.dev_mode_size, random_state=SEED)
data = {
'input': {
'meta': meta_test,
},
'callback_input': {
'meta_valid': None
}
}
pipeline_network = unet(config=CONFIG, train_mode=False)
pipeline_postprocessing = pipelines.mask_postprocessing(config=CONFIG)
pipeline_network.clean_cache()
predicted_masks = pipeline_network.transform(data)
test_masks = {'input_masks': predicted_masks}
output = pipeline_postprocessing.transform(test_masks)
pipeline_network.clean_cache()
pipeline_postprocessing.clean_cache()
y_pred_test = output['binarized_images']
submission = utils.create_submission(meta_test, y_pred_test)
submission_filepath = os.path.join(EXPERIMENT_DIR, 'submission.csv')
submission.to_csv(submission_filepath, index=None, encoding='utf-8')
LOGGER.info('submission saved to {}'.format(submission_filepath))
LOGGER.info('submission head \n\n{}'.format(submission.head()))
def fold_evaluate_loop(valid_data_split, fold_id):
valid_pipe_input = {
'input': {
'meta': valid_data_split
},
'callback_input': {
'meta_valid': None
}
}
valid_ids = valid_data_split[ID_COLUMN].tolist()
LOGGER.info('Start pipeline transform on valid')
pipeline_network = unet(config=CONFIG,
suffix='_fold_{}'.format(fold_id),
train_mode=False)
pipeline_postprocessing = pipelines.mask_postprocessing(
config=CONFIG, suffix='_fold_{}'.format(fold_id))
pipeline_network.clean_cache()
pipeline_postprocessing.clean_cache()
predicted_masks_valid = pipeline_network.transform(valid_pipe_input)
valid_pipe_masks = {'input_masks': predicted_masks_valid}
output_valid = pipeline_postprocessing.transform(valid_pipe_masks)
utils.clean_object_from_memory(pipeline_network)
y_pred_valid = output_valid['binarized_images']
y_true_valid = utils.read_masks(valid_data_split[Y_COLUMN].values)
iou, iout = calculate_scores(y_true_valid, y_pred_valid)
predicted_masks_valid = predicted_masks_valid['resized_images']
return iou, iout, (valid_ids, predicted_masks_valid)
def evaluate():
meta = pd.read_csv(PARAMS.metadata_filepath)
meta_train = meta[meta['is_train'] == 1]
cv = utils.KFoldBySortedValue(n_splits=PARAMS.n_cv_splits,
shuffle=PARAMS.shuffle,
random_state=SEED)
for train_idx, valid_idx in cv.split(
meta_train[DEPTH_COLUMN].values.reshape(-1)):
break
meta_valid_split = meta_train.iloc[valid_idx]
y_true_valid = utils.read_masks(meta_valid_split[Y_COLUMN].values)
if DEV_MODE:
meta_valid_split = meta_valid_split.sample(PARAMS.dev_mode_size,
random_state=SEED)
data = {
'input': {
'meta': meta_valid_split,
},
'callback_input': {
'meta_valid': None
}
}
pipeline_network = unet(config=CONFIG, train_mode=False)
pipeline_postprocessing = pipelines.mask_postprocessing(config=CONFIG)
pipeline_network.clean_cache()
output = pipeline_network.transform(data)
valid_masks = {'input_masks': output}
output = pipeline_postprocessing.transform(valid_masks)
pipeline_network.clean_cache()
pipeline_postprocessing.clean_cache()
y_pred_valid = output['binarized_images']
LOGGER.info('Calculating IOU and IOUT Scores')
iou_score, iout_score = calculate_scores(y_true_valid, y_pred_valid)
LOGGER.info('IOU score on validation is {}'.format(iou_score))
CTX.channel_send('IOU', 0, iou_score)
LOGGER.info('IOUT score on validation is {}'.format(iout_score))
CTX.channel_send('IOUT', 0, iout_score)
results_filepath = os.path.join(EXPERIMENT_DIR, 'validation_results.pkl')
LOGGER.info('Saving validation results to {}'.format(results_filepath))
joblib.dump((meta_valid_split, y_true_valid, y_pred_valid),
results_filepath)