def predict_dataset(self, dataset, progress_bar=True, apply_preproc=True):
""" Predict a complete dataset
Parameters
----------
dataset : Dataset
Dataset to predict
progress_bar : bool, optional
hide or show a progress bar
Yields
-------
PredictionResult
Single PredictionResult
dict
Dataset entry of the prediction result
"""
if isinstance(dataset, RawDataSet):
input_dataset = StreamingInputDataset(dataset, self.data_preproc if apply_preproc else None, self.text_postproc if apply_preproc else None)
else:
input_dataset = RawInputDataset(dataset, self.data_preproc if apply_preproc else None, self.text_postproc if apply_preproc else None)
prediction_results = self.predict_input_dataset(input_dataset, progress_bar)
for prediction, sample in zip(prediction_results, dataset.samples()):
yield prediction, sample
def predict_dataset(self, dataset, progress_bar=True):
start_time = time.time()
with StreamingInputDataset(
dataset,
self.predictors[0].data_preproc,
self.predictors[0].text_postproc,
None,
processes=self.processes,
) as input_dataset:
def progress_bar_wrapper(l):
if progress_bar:
return tqdm(l,
total=int(
np.ceil(len(dataset) / self.batch_size)),
desc="Prediction")
else:
return l
def batched_data_params():
batch = []
for data_idx, (image, _, params) in enumerate(
input_dataset.generator(epochs=1)):
batch.append((data_idx, image, params))
if len(batch) == self.batch_size:
yield batch
batch = []
if len(batch) > 0:
yield batch
for batch in progress_bar_wrapper(batched_data_params()):
sample_ids, batch_images, batch_params = zip(*batch)
samples = [dataset.samples()[i] for i in sample_ids]
current_mode = dataset.mode
with ExitStack() as stack:
raw_dataset = [
stack.enter_context(
RawInputDataset(
current_mode,
batch_images,
[None] * len(batch_images),
batch_params,
)) for _ in self.predictors
]
# predict_raw returns list of prediction objects
prediction = [
predictor.predict_input_dataset(ds, progress_bar=False)
for ds, predictor in zip(raw_dataset, self.predictors)
]
for result, sample in zip(zip(*prediction), samples):
yield result, sample
print("Prediction of {} models took {}s".format(
len(self.predictors),
time.time() - start_time))
def predict_dataset(self, dataset, progress_bar=True):
start_time = time.time()
dataset.load_samples(processes=1, progress_bar=progress_bar)
datas = dataset.prediction_samples()
# preprocessing step (if all share the same preprocessor)
if self.same_preproc:
data_params = self.predictors[0].data_preproc.apply(
datas, processes=self.processes, progress_bar=progress_bar)
else:
raise Exception(
'Different preprocessors are currently not allowed during prediction'
)
def progress_bar_wrapper(l):
if progress_bar:
l = list(l)
return tqdm(l, total=len(l), desc="Prediction")
else:
return l
for data_idx in progress_bar_wrapper(
range(0, len(datas), self.batch_size)):
batch_data_params = data_params[data_idx:data_idx +
self.batch_size]
samples = dataset.samples()[data_idx:data_idx + self.batch_size]
raw_dataset = [
RawInputDataset(
DataSetMode.PREDICT,
[img for img, _ in batch_data_params],
[None] * len(batch_data_params),
[p for _, p in batch_data_params],
None if self.same_preproc else p.data_preproc,
None if self.same_preproc else p.text_postproc,
) for p in self.predictors
]
# predict_raw returns list of prediction objects
prediction = [
predictor.predict_input_dataset(ds, progress_bar=False)
for ds, predictor in zip(raw_dataset, self.predictors)
]
for result, sample in zip(zip(*prediction), samples):
yield result, sample
print("Prediction of {} models took {}s".format(
len(self.predictors),
time.time() - start_time))
def predict_dataset(self, dataset, progress_bar=True):
start_time = time.time()
# preprocessing step (if all share the same preprocessor)
if not self.same_preproc:
raise Exception(
'Different preprocessors are currently not allowed during prediction'
)
input_dataset = InputDataset(
dataset,
self.predictors[0].data_preproc,
self.predictors[0].text_postproc,
None,
processes=self.processes,
)
def progress_bar_wrapper(l):
if progress_bar:
return tqdm(l,
total=int(np.ceil(len(dataset) / self.batch_size)),
desc="Prediction")
else:
return l
def batched_data_params():
batch = []
for data_idx, (image, _, params) in enumerate(
input_dataset.generator(epochs=1)):
batch.append((data_idx, image, params))
if len(batch) == self.batch_size:
yield batch
batch = []
if len(batch) > 0:
yield batch
for batch in progress_bar_wrapper(batched_data_params()):
sample_ids, batch_images, batch_params = zip(*batch)
samples = [dataset.samples()[i] for i in sample_ids]
raw_dataset = [
RawInputDataset(
DataSetMode.PREDICT,
batch_images,
[None] * len(batch_images),
batch_params,
None,
None,
) for p in self.predictors
]
# predict_raw returns list of prediction objects
prediction = [
predictor.predict_input_dataset(ds, progress_bar=False)
for ds, predictor in zip(raw_dataset, self.predictors)
]
for result, sample in zip(zip(*prediction), samples):
yield result, sample
print("Prediction of {} models took {}s".format(
len(self.predictors),
time.time() - start_time))