2025-08-11 12:24:21 +08:00

219 lines
9.3 KiB
Python

# Copyright 2020 The MediaPipe Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""Tests for mediapipe.python.solutions.hands."""
import json
import os
import tempfile # pylint: disable=unused-import
from typing import NamedTuple
from absl.testing import absltest
from absl.testing import parameterized
import cv2
import numpy as np
import numpy.testing as npt
# resources dependency
# undeclared dependency
from mediapipe.python.solutions import drawing_styles
from mediapipe.python.solutions import drawing_utils as mp_drawing
from mediapipe.python.solutions import hands as mp_hands
TEST_IMAGE_PATH = 'mediapipe/python/solutions/testdata'
LITE_MODEL_DIFF_THRESHOLD = 25 # pixels
FULL_MODEL_DIFF_THRESHOLD = 20 # pixels
EXPECTED_HAND_COORDINATES_PREDICTION = [[[580, 34], [504, 50], [459, 94],
[429, 146], [397, 182], [507, 167],
[479, 245], [469, 292], [464, 330],
[545, 180], [534, 265], [533, 319],
[536, 360], [581, 172], [587, 252],
[593, 304], [599, 346], [615, 168],
[628, 223], [638, 258], [648, 288]],
[[138, 343], [211, 330], [257, 286],
[289, 237], [322, 203], [219, 216],
[238, 138], [249, 90], [253, 51],
[177, 204], [184, 115], [187, 60],
[185, 19], [138, 208], [131, 127],
[124, 77], [117, 36], [106, 222],
[92, 159], [79, 124], [68, 93]]]
class HandsTest(parameterized.TestCase):
def _get_output_path(self, name):
return os.path.join(tempfile.gettempdir(), self.id().split('.')[-1] + name)
def _landmarks_list_to_array(self, landmark_list, image_shape):
rows, cols, _ = image_shape
return np.asarray([(lmk.x * cols, lmk.y * rows, lmk.z * cols)
for lmk in landmark_list.landmark])
def _world_landmarks_list_to_array(self, landmark_list):
return np.asarray([(lmk.x, lmk.y, lmk.z)
for lmk in landmark_list.landmark])
def _assert_diff_less(self, array1, array2, threshold):
npt.assert_array_less(np.abs(array1 - array2), threshold)
def _annotate(self, frame: np.ndarray, results: NamedTuple, idx: int):
for hand_landmarks in results.multi_hand_landmarks:
mp_drawing.draw_landmarks(
frame, hand_landmarks, mp_hands.HAND_CONNECTIONS,
drawing_styles.get_default_hand_landmarks_style(),
drawing_styles.get_default_hand_connections_style())
path = os.path.join(tempfile.gettempdir(), self.id().split('.')[-1] +
'_frame_{}.png'.format(idx))
cv2.imwrite(path, frame)
def test_invalid_image_shape(self):
with mp_hands.Hands() as hands:
with self.assertRaisesRegex(
ValueError, 'Input image must contain three channel rgb data.'):
hands.process(np.arange(36, dtype=np.uint8).reshape(3, 3, 4))
def test_blank_image(self):
with mp_hands.Hands() as hands:
image = np.zeros([100, 100, 3], dtype=np.uint8)
image.fill(255)
results = hands.process(image)
self.assertIsNone(results.multi_hand_landmarks)
self.assertIsNone(results.multi_handedness)
@parameterized.named_parameters(
('static_image_mode_with_lite_model', True, 0, 5),
('video_mode_with_lite_model', False, 0, 10),
('static_image_mode_with_full_model', True, 1, 5),
('video_mode_with_full_model', False, 1, 10))
def test_multi_hands(self, static_image_mode, model_complexity, num_frames):
image_path = os.path.join(os.path.dirname(__file__), 'testdata/hands.jpg')
image = cv2.imread(image_path)
with mp_hands.Hands(
static_image_mode=static_image_mode,
max_num_hands=2,
model_complexity=model_complexity,
min_detection_confidence=0.5) as hands:
for idx in range(num_frames):
results = hands.process(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))
self._annotate(image.copy(), results, idx)
handedness = [
handedness.classification[0].label
for handedness in results.multi_handedness
]
multi_hand_coordinates = []
rows, cols, _ = image.shape
for landmarks in results.multi_hand_landmarks:
self.assertLen(landmarks.landmark, 21)
x = [landmark.x * cols for landmark in landmarks.landmark]
y = [landmark.y * rows for landmark in landmarks.landmark]
hand_coordinates = np.column_stack((x, y))
multi_hand_coordinates.append(hand_coordinates)
self.assertLen(handedness, 2)
self.assertLen(multi_hand_coordinates, 2)
prediction_error = np.abs(
np.asarray(multi_hand_coordinates) -
np.asarray(EXPECTED_HAND_COORDINATES_PREDICTION))
diff_threshold = LITE_MODEL_DIFF_THRESHOLD if model_complexity == 0 else FULL_MODEL_DIFF_THRESHOLD
npt.assert_array_less(prediction_error, diff_threshold)
def _process_video(self, model_complexity, video_path,
max_num_hands=1,
num_landmarks=21,
num_dimensions=3):
# Predict pose landmarks for each frame.
video_cap = cv2.VideoCapture(video_path)
landmarks_per_frame = []
w_landmarks_per_frame = []
with mp_hands.Hands(
static_image_mode=False,
max_num_hands=max_num_hands,
model_complexity=model_complexity,
min_detection_confidence=0.5) as hands:
while True:
# Get next frame of the video.
success, input_frame = video_cap.read()
if not success:
break
# Run pose tracker.
input_frame = cv2.cvtColor(input_frame, cv2.COLOR_BGR2RGB)
frame_shape = input_frame.shape
result = hands.process(image=input_frame)
frame_landmarks = np.zeros([max_num_hands,
num_landmarks, num_dimensions]) * np.nan
frame_w_landmarks = np.zeros([max_num_hands,
num_landmarks, num_dimensions]) * np.nan
if result.multi_hand_landmarks:
for idx, landmarks in enumerate(result.multi_hand_landmarks):
landmarks = self._landmarks_list_to_array(landmarks, frame_shape)
frame_landmarks[idx] = landmarks
if result.multi_hand_world_landmarks:
for idx, w_landmarks in enumerate(result.multi_hand_world_landmarks):
w_landmarks = self._world_landmarks_list_to_array(w_landmarks)
frame_w_landmarks[idx] = w_landmarks
landmarks_per_frame.append(frame_landmarks)
w_landmarks_per_frame.append(frame_w_landmarks)
return (np.array(landmarks_per_frame), np.array(w_landmarks_per_frame))
@parameterized.named_parameters(
('full', 1, 'asl_hand.full.npz'))
def test_on_video(self, model_complexity, expected_name):
"""Tests hand models on a video."""
# Set threshold for comparing actual and expected predictions in pixels.
diff_threshold = 18
world_diff_threshold = 0.05
video_path = os.path.join(os.path.dirname(__file__),
'testdata/asl_hand.25fps.mp4')
expected_path = os.path.join(os.path.dirname(__file__),
'testdata/{}'.format(expected_name))
actual, actual_world = self._process_video(model_complexity, video_path)
# Dump actual .npz.
npz_path = self._get_output_path(expected_name)
np.savez(npz_path, predictions=actual, w_predictions=actual_world)
# Dump actual JSON.
json_path = self._get_output_path(expected_name.replace('.npz', '.json'))
with open(json_path, 'w') as fl:
dump_data = {
'predictions': np.around(actual, 3).tolist(),
'predictions_world': np.around(actual_world, 3).tolist()
}
fl.write(json.dumps(dump_data, indent=2, separators=(',', ': ')))
# Validate actual vs. expected landmarks.
expected = np.load(expected_path)['predictions']
assert actual.shape == expected.shape, (
'Unexpected shape of predictions: {} instead of {}'.format(
actual.shape, expected.shape))
self._assert_diff_less(
actual[..., :2], expected[..., :2], threshold=diff_threshold)
# Validate actual vs. expected world landmarks.
expected_world = np.load(expected_path)['w_predictions']
assert actual_world.shape == expected_world.shape, (
'Unexpected shape of world predictions: {} instead of {}'.format(
actual_world.shape, expected_world.shape))
self._assert_diff_less(
actual_world, expected_world, threshold=world_diff_threshold)
if __name__ == '__main__':
absltest.main()