def preprocess(self, data):
preprocessed_data = []
data_infos = []
for img in data:
img = np.array(img)[:, :, ::-1]
height, width, _ = img.shape
if self.input_height / self.input_width < height / width:
scale = self.input_height / height
else:
scale = self.input_width / width
scaled_img = cv2.resize(
img, (0, 0), fx=scale, fy=scale, interpolation=cv2.INTER_CUBIC,
)
padded_img, pad = pad_img(
scaled_img, (0, 0, 0), [self.input_height, self.input_width],
)
padded_img = padded_img.transpose((2, 0, 1))
padded_img = padded_img[np.newaxis].astype(np.float32)
preprocessed_data.append(padded_img)
data_infos.append(
DataInfo(
scales=(scale, scale),
pads=tuple(pad),
original_width=width,
original_height=height,
),
)
return [preprocessed_data, data_infos]
def forward(self, in_queue, out_queue, data):
results = []
if data is not None:
for sample in data[0]:
nn_data = dai.NNData()
nn_data.setLayer("data", sample)
in_queue.send(nn_data)
assert wait_for_results(out_queue)
results.append(out_queue.get())
data[0] = results
else:
assert wait_for_results(out_queue)
results.append(out_queue.get())
data = [
results,
[
DataInfo(
scales=(
self.input_width / self.video_width,
self.input_height / self.video_height,
),
pads=(0, 0),
original_width=self.video_width,
original_height=self.video_height,
),
],
]
return data
def preprocess(self, data):
preprocessed_data = []
data_infos = []
for img in data:
img = np.array(img)
height, width, _ = img.shape
scale_y = self.input_height / height
scale_x = self.input_width / width
resized_image = cv2.resize(
img,
(self.input_width, self.input_height),
interpolation=cv2.INTER_CUBIC,
)
self.resized_image = resized_image
resized_image = resized_image.transpose((2, 0, 1))
resized_image = resized_image[np.newaxis].astype(np.float32)
preprocessed_data.append(resized_image)
data_infos.append(
DataInfo(
scales=(scale_y, scale_x),
pads=(0, 0),
original_width=width,
original_height=height,
), )
return [preprocessed_data, data_infos]
def preprocess(self, data):
preprocessed_data = []
data_infos = []
for img in data:
img = np.array(img)
height, width, _ = img.shape
scale_x = self.input_width / width
scale_y = self.input_height / height
scaled_img = cv2.resize(img, (self.input_height, self.input_width))
scaled_img = scaled_img.astype(np.float32)
scaled_img -= self.mean
scaled_img = scaled_img.transpose((2, 0, 1))
scaled_img = scaled_img[np.newaxis]
data_infos.append(
DataInfo(
scales=(scale_x, scale_y),
pads=(0, 0),
original_width=width,
original_height=height,
),
)
preprocessed_data.append(scaled_img)
return [preprocessed_data, data_infos]