def test_get_top_k_without_responses(self):
"""Tests the selection of top entries in a keypoints w/o responses."""
input_keypoints = Keypoints(
coordinates=np.array(
[
[10.0, 23.2],
[37.1, 50.2],
[90.1, 10.7],
[150.0, 122.0],
[250.0, 49.0],
]
)
)
# test with requested length > current length
requested_length = len(input_keypoints) * 2
computed = input_keypoints.get_top_k(requested_length)
self.assertEqual(computed, input_keypoints)
# test with requested length < current length
requested_length = 2
computed = input_keypoints.get_top_k(requested_length)
expected = Keypoints(coordinates=input_keypoints.coordinates[:requested_length])
self.assertEqual(computed, expected)
def detect_and_describe(self,
image: Image) -> Tuple[Keypoints, np.ndarray]:
"""Jointly generate keypoint detections and their associated descriptors from a single image."""
# TODO(ayushbaid): fix inference issue #110
device = torch.device("cuda" if self._use_cuda else "cpu")
model = SuperPoint(self._config).to(device)
model.eval()
# Compute features.
image_tensor = torch.from_numpy(
np.expand_dims(
image_utils.rgb_to_gray_cv(image).value_array.astype(
np.float32) / 255.0, (0, 1))).to(device)
with torch.no_grad():
model_results = model({"image": image_tensor})
torch.cuda.empty_cache()
# Unpack results.
coordinates = model_results["keypoints"][0].detach().cpu().numpy()
scores = model_results["scores"][0].detach().cpu().numpy()
keypoints = Keypoints(coordinates, scales=None, responses=scores)
descriptors = model_results["descriptors"][0].detach().cpu().numpy().T
# Filter features.
if image.mask is not None:
keypoints, valid_idxs = keypoints.filter_by_mask(image.mask)
descriptors = descriptors[valid_idxs]
keypoints, selection_idxs = keypoints.get_top_k(self.max_keypoints)
descriptors = descriptors[selection_idxs]
return keypoints, descriptors
def test_get_top_k_with_responses(self):
"""Tests the selection of top entries in a keypoints with responses."""
input_keypoints = Keypoints(
coordinates=np.array(
[
[10.0, 23.2],
[37.1, 50.2],
[90.1, 10.7],
[150.0, 122.0],
[250.0, 49.0],
]
),
scales=np.array([1, 3, 2, 3.2, 1.8]),
responses=np.array([0.3, 0.7, 0.9, 0.1, 0.2]),
)
# test with requested length > current length
requested_length = len(input_keypoints) * 2
computed = input_keypoints.get_top_k(requested_length)
self.assertEqual(computed, input_keypoints)
# test with requested length < current length
requested_length = 2
computed = input_keypoints.get_top_k(requested_length)
expected = Keypoints(
coordinates=np.array(
[
[37.1, 50.2],
[90.1, 10.7],
]
),
scales=np.array([3, 2]),
responses=np.array([0.7, 0.9]),
)
# compare in an order-insensitive fashion
self.compare_without_ordering(computed, expected)