def test_multicamgrablabeler():
"""Generate simulate data_collection data and run calibrator."""
import infer_sequence
base_dir = 'data/batch2/test_data'
labeler = MultiCamGrabLabeler('data/batch3')
for i in range(0, 13):
filename = str(10000000 + i + 1)[1:] + '.jpg'
cam_data = {
'cam1_color_file':
os.path.join(base_dir, '4', filename),
'cam1_dept_file':
os.path.join(base_dir, '6', filename),
'cam2_color_file':
os.path.join(base_dir, '5', filename),
'cam2_dept_file':
os.path.join(base_dir, '7', filename),
'xeye_image': [
os.path.join(base_dir, '0', filename),
os.path.join(base_dir, '1', filename),
os.path.join(base_dir, '2', filename),
os.path.join(base_dir, '3', filename),
]
}
labeler.add_files(cam_data)
results = labeler.gen_label()
if not results:
continue
images = []
for r in results:
images.extend(r['images'])
infer_sequence.show_multi_images(images, resize_ratio=0.5, cols=4)
def rerun_multicamgrablabeler(base_dir, num):
"""Generate simulate data_collection data and run calibrator."""
import infer_sequence
# base_dir = 'data/batch4'
collection_dir = 'test_data'
labeler = MultiCamGrabLabeler(base_dir).init_from_test_data(collection_dir)
num = int(num)
for i in range(2, num):
results = labeler.gen_label(counter=i)
images = []
for r in results:
images.extend(r['images'])
infer_sequence.show_multi_images(images, resize_ratio=0.5, cols=2)
def convert_to_airupdate(base_dir, num, output_arg):
from align_all_sensor import MultiCamGrabLabeler
import infer_sequence
from kvtools import append_kv
from kvtools import convert_to_stream_out
from attrs_utils import to_bytes
# base_dir = 'data/batch4'
collection_dir = 'test_data'
labeler = MultiCamGrabLabeler(base_dir).init_from_test_data(collection_dir)
out = convert_to_stream_out(output_arg)
num = int(num)
for i in range(2, num):
results = labeler.gen_label(counter=i)
images = []
for r in results:
images.extend(r['images'])
infer_sequence.show_multi_images(images, resize_ratio=0.5, cols=2)
f = open(
os.path.join(base_dir, 'test_data', '0',
str(10000000 + i)[1:] + '.txt'))
product_id = f.read().strip()
for res in results:
for r in res['combined']:
print('r:', r.keys())
x, y, degree = r['rot_rect']
start_frame = AirProductFrame(filepath=r['rgb0'],
data=open(r['rgb0'],
'rb').read())
end_frame = AirProductFrame(filepath=r['rgb1'],
data=open(r['rgb1'], 'rb').read())
rs_info = RealsenseAssistInfo(
poly=r['poly'],
sum_depth_diff=r['sum_depth_diff'],
rect=r['rect'],
rot_rect=(x[0], x[1], y[0], y[1], degree),
region=r['region'],
contour=r['contour'])
action = 1 if r['sum_depth_diff'] > 0 else -1
record_id = base_dir + '_' + str(i)
airup = AirRecordOfBuildSearchUpdate(start_frame=start_frame,
end_frame=end_frame,
goods_num=action,
action=action,
product_id=product_id,
record_id=record_id,
key=record_id,
rs_assist_info=rs_info)
v = to_bytes(airup)
append_kv(airup.key.encode('utf-8'), v, out)
def show_images(self, resize_ratio=0.5, cols=4):
import cv2
import infer_sequence
print(self.counter)
for i in range(self.counter):
images = []
for j in self.rgb_cam_list:
n = str(10000000 + i + 1)[1:]
path = os.path.join(self.calib_data_dir, str(j), 'cam0', n)
if not os.path.exists(path + '.jpg'):
path = path + '.png'
else:
path = path + '.jpg'
print(path)
images.append(cv2.imread(path))
infer_sequence.show_multi_images(images, resize_ratio, cols)
def show_images(self, images, resize_ratio=0.5, cols=2):
import infer_sequence
logger.info('show image. image size: {}'.format(len(images)))
return infer_sequence.show_multi_images(images, resize_ratio, cols)