def preprocess_frame(data,
output_dim,
use_thermal=True,
augment=False,
preprocess_fn=None):
if use_thermal:
channel = TrackChannels.thermal
else:
channel = TrackChannels.filtered
data = data[channel]
max = np.amax(data)
min = np.amin(data)
if max == min:
return None
data -= min
data = data / (max - min)
np.clip(data, a_min=0, a_max=None, out=data)
data = data[np.newaxis, :]
data = np.transpose(data, (1, 2, 0))
data = np.repeat(data, output_dim[2], axis=2)
data = imageprocessing.resize_cv(data, output_dim)
# preprocess expects values in range 0-255
if preprocess_fn:
data = data * 255
data = preprocess_fn(data)
return data
def resize_frame(frame, channel, frame_size, keep_aspect=False):
if keep_aspect:
return imageprocessing.resize_with_aspect(
frame, (frame_size, frame_size), channel
)
return imageprocessing.resize_cv(
np.float32(frame), (frame_size, frame_size), channel
)
def augement_frame(frame, frame_size, dim):
frame = imageprocessing.resize_cv(
frame,
dim,
extra_h=random.randint(0, int(frame_size * 0.05)),
extra_v=random.randint(0, int(frame_size * 0.05)),
)
image = convert(frame)
image = tf.image.random_crop(image, size=[dim[0], dim[1], 3])
if random.random() > 0.50:
image = tf.image.flip_left_right(image)
if random.random() > 0.20:
image = tf.image.random_contrast(image, 0.8, 1.2)
image = tf.minimum(image, 1.0)
image = tf.maximum(image, 0.0)
return image.numpy()
def preprocess_frame(
data, output_dim, use_thermal=True, augment=False, preprocess_fn=None
):
if use_thermal:
channel = TrackChannels.thermal
else:
channel = TrackChannels.filtered
data = data[channel]
data, stats = imageprocessing.normalize(data)
if not stats[0]:
return None
data = data[np.newaxis, :]
data = np.transpose(data, (1, 2, 0))
data = np.repeat(data, output_dim[2], axis=2)
data = imageprocessing.resize_cv(data, output_dim, channel)
# preprocess expects values in range 0-255
if preprocess_fn:
data = data * 255
data = preprocess_fn(data)
return data
def resize(self, dim):
self.thermal = resize_cv(self.thermal, dim)
self.mask = resize_cv(self.mask, dim, interpolation=cv2.INTER_NEAREST)
if self.flow is not None:
self.flow = resize_cv(self.flow, dim)
self.filtered = resize_cv(self.filtered, dim)