def set_params_disp(disp='cnn'):
params = {}
params['data_path'] = HOME + 'kitti/2012/displets_data/displets_data/Kitti/training/'
params['image_path'] = params['data_path'] + 'colored_0/'
params['gt_disp_path'] = params['data_path'] + 'gt/disp_noc/'
params['cnn_disp_path'] = params['data_path'] + 'dispmaps/%s/disp_0/' % disp
params['sem_label_path'] = HOME + 'kitti/2012/displets_data/displets_semantic_labels/'
params['label_path'] = params['sem_label_path'] + 'semantic/'
params['label_instance_path'] = params['sem_label_path'] + 'instance/'
params['car_instance_path'] = params['sem_label_path'] + 'car_instance/'
params['car_inst_path'] = params['sem_label_path'] + 'car_inst_denoise/'
uts.mkdir_if_need(params['car_inst_path'])
params['output_path'] = params['data_path'] + 'Results/'
params['car_model_path'] = HOME + 'kitti/2012/displets_data/displets_data/models/semi_convex_hull/'
car_ids = [1, 7, 8, 9, 10, 11, 13, 15]
params['car_names'] = ['%02d' % x for x in car_ids]
params['car_num'] = len(car_ids)
params['car_inst_num'] = 652
params['image_set'] = params['data_path'] + 'split/train.txt'
params['calib_path'] = params['data_path'] + 'calib/'
params['intrinsic'] = []
color_params = zpark.gen_color_list(HOME + 'zpark/color_v2.lst')
params['size'] = [375, 1242]
params.update(color_params)
params['colors'] = {'car': [64, 0, 128]}
params['class_names'] = ['car']
return params
def event_handler(event):
if isinstance(event, paddle.event.EndIteration):
if event.batch_id % 50 == 0:
print "\nPass %d, Batch %d, Cost %f" % (
event.pass_id, event.batch_id, event.cost)
else:
sys.stdout.write('.')
sys.stdout.flush()
elif isinstance(event, paddle.event.EndPass):
result = trainer.test(reader=paddle.batch(
sun3d.test(params['test_scene'][0:4],
height=params['size'][0],
width=params['size'][1],
tasks=tasks),
batch_size=2 * params['batch_size']),
feeding=feeding)
task_string = '_'.join(tasks)
print "\nTask %s, Pass %d, Cost %f" % (task_string, event.pass_id,
result.cost)
folder = params['output_path'] + '/' + data_source
uts.mkdir_if_need(folder)
model_name = folder + '/model_stage_' + str(
params['stage']) + '_' + task_string + '.tar.gz'
with gzip.open(model_name, 'w') as f:
parameters.to_tar(f)
print "\nsave with pass %d" % (event.pass_id)
def event_handler(event):
if isinstance(event, paddle.event.EndIteration):
if event.batch_id % 50 == 0:
if not isinstance(event.cost, list):
cost = [event.cost]
else:
cost = event.cost
print "\nPass %d, Batch %d, " % (event.pass_id,
event.batch_id),
for i in range(len(cost)):
print "%s: %f, " % (cost_name[i], cost[i]),
print "\n"
else:
sys.stdout.write('.')
sys.stdout.flush()
elif (isinstance(event, paddle.event.EndPass) and \
event.pass_id % 4 == 1):
print "Testing",
result = trainer.test(reader=paddle.batch(
sun3d.test_upsampler(params['test_scene'][0:5],
rate=params['sample_rate'],
height=params['size'][0],
width=params['size'][1]),
batch_size=params['batch_size']),
feeding=feeding)
print "\nTask upsample, Pass %d," % (event.pass_id),
if not isinstance(result.cost, list):
cost = [result.cost]
else:
cost = result.cost
for i in range(len(cost)):
print "%s: %f, " % (cost_name[i], cost[i]),
folder = params['output_path'] + '/upsampler/'
uts.mkdir_if_need(folder)
model_name = folder + '/upsample_model_' + \
FLAGS.suffix + '.tar.gz'
with gzip.open(model_name, 'w') as f:
parameters.to_tar(f)
print "model saved at %s" % model_name
def main(argv):
argv = FLAGS(argv) # parse argv to FLAG
scene_names = uts.read_file(DATA_PATH + 'test_sun3d.txt')
scene_names = scene_names[FLAGS.start:FLAGS.end]
for scene_name in scene_names:
scene_name += '/'
print FLOW_PATH + scene_name + 'flow/'
# down load file
if not os.path.exists(DATA_PATH + scene_name + 'id_img2depth.txt'):
print 'Retrieve Data'
os.system("/home/peng/SUN3DCppReader/src/build/SUN3DCppReader " \
+ scene_name + " " + DATA_PATH)
if not os.path.exists(DATA_PATH + scene_name +
'/id_img2depth.txt'):
continue
if not os.path.exists(FLOW_PATH + scene_name + '/flow/'):
uts.mkdir_if_need(FLOW_PATH + scene_name + '/flow/')
id_img2depth = find_image_depth_matching(scene_name)
gen_img_pair_data(scene_name, 500, id_img2depth)
def set_params_disp(disp='psm', stereo_rect=True):
params = {}
params['data_path'] = HOME + 'apolloscape/car_labels/'
params['image_path'] = params['data_path'] + 'images/'
params['depth_gt_path'] = params['data_path'] + 'depth/'
params['depth_path_rect'] = params['data_path'] + 'stereo_depth/%s/' % disp
params['depth_path_v2'] = params['data_path'] + 'depth_2/'
params['car_pose_path'] = params['data_path'] + 'car_poses/'
params['car_pose_path_new'] = params['data_path'] + 'car_poses_challenge/'
params[
'car_pose_path_sim'] = params['data_path'] + 'car_poses_challenge_sim/'
uts.mkdir_if_need(params['car_pose_path_new'])
uts.mkdir_if_need(params['car_pose_path_sim'])
params['car_inst_path'] = params['data_path'] + 'car_instance/'
params['train_list'] = params['data_path'] + 'split/train.txt'
params['val_list'] = params['data_path'] + 'split/val.txt'
params['minival_list'] = params['data_path'] + 'split/mini_val.txt'
params['minitrain_list'] = params['data_path'] + 'split/mini_train.txt'
params['output_path'] = params['data_path'] + 'results/'
params['car_inst_num'] = 10000
params['size'] = [2710, 3384]
params['intrinsic'] = {
'Camera_5':
np.array([
2304.54786556982, 2305.875668062, 1686.23787612802,
1354.98486439791
]),
'Camera_6':
np.array([
2300.39065314361, 2301.31478860597, 1713.21615190657,
1342.91100799715
])
}
# normalized intrinsic
cam_names = params['intrinsic'].keys()
for c_name in cam_names:
params['intrinsic'][c_name][[0, 2]] /= params['size'][1]
params['intrinsic'][c_name][[1, 3]] /= params['size'][0]
vertice_num = '5000'
params['car_model_path'] = HOME + 'car_models/%s/' % vertice_num
params['car_model_path'] = HOME + 'car_models/5000_align/'
params['car_model_path_pkl'] = params[
'data_path'] + '/3d_car_instance_sample/car_models/'
params['car_model_path_off'] = HOME + '/car_models/%s_off/' % vertice_num
params['car_model_path_pkl'] = HOME + '/car_models/%s_pkl/' % vertice_num
params['save_image_path'] = params[
'data_path'] + '/3d_car_instance_sample/images/'
# params['car_model_path'] = HOME + 'car_models/5000/'
# params['car_model_path_pkl'] = HOME + 'car_models/5000_pkl/'
uts.mkdir_if_need(params['car_model_path_pkl'])
uts.mkdir_if_need(params['car_model_path_off'])
# params['car_names'] = [os.path.basename(x)[:-4] for x in \
# glob.glob(params['car_model_path'] + '*.mat')]
params['car_types'], params['car_names'] = load_carnames(
params['car_model_path'])
color_params = zpark.gen_color_list(HOME + 'zpark/color_v2.lst')
params.update(color_params)
if stereo_rect:
# the crop with cars
params['rect_crop'] = np.array([1432., 668., 2200., 2716.])
params['rect_crop'][[0, 2]] /= params['size'][0]
params['rect_crop'][[1, 3]] /= params['size'][1]
# the crop only crop vertically
params['rect_crop_top'] = np.array([1432., 0., 2200., 3383.])
params['rect_crop_top'][[0, 2]] /= params['size'][0]
params['rect_crop_top'][[1, 3]] /= params['size'][1]
params['image_path_rect'] = params['image_path'] + 'stereo_rect/'
uts.mkdir_if_need(params['image_path_rect'])
params['car_inst_path_rect'] = params['car_inst_path'] + 'stereo_rect/'
uts.mkdir_if_need(params['car_inst_path_rect'])
rectify_params = get_stereo_rect_extern(params)
params.update(rectify_params)
return params
def sequencial_upsampleing(dataset='sun3d',
split='train',
max_num=None,
vis=False):
# Read image pair 1, 2, generate depth
if dataset == 'sun3d':
params = sun3d.set_params()
params['demon_model'] = '../output/tf_model_full_5.tar.gz'
else:
print "dataset {} is not supported".format(dataset)
deep_upsampler = DeepUpSampler(params)
part, part_id = [int(x) for x in FLAGS.part.split(',')]
test_ids = partition(len(params[split + '_scene']), part, part_id)
rate = 0.05
process_scene_names = [params[split + '_scene'][x] for x in test_ids]
all_time = 0.
all_count = 0.
for scene_name in process_scene_names:
image_list = preprocess_util.list_files(params['flow_path'] +
scene_name + '/flow/')
image2depth = sun3d.get_image_depth_matching(scene_name)
image_num = len(image_list) if max_num is None \
else min(len(image_list), max_num)
image_id = range(0, len(image_list), len(image_list) / image_num)
upsample_output_path = params['flow_path'] + scene_name + \
'/pair_depth/' + str(rate) + '/'
uts.mkdir_if_need(upsample_output_path)
print "processing {} with images: {}".format(scene_name, len(image_id))
image_name_list = [image_list[x] for x in image_id]
for pair_name in image_name_list:
pair_image_name = pair_name.split('/')[-1]
outfile = upsample_output_path + pair_image_name[:-4] + '.npy'
# if uts.exists(outfile):
# print "\t {} exists".format(pair_name)
# continue
image1, image2, flow_gt, depth_gt = \
sun3d.load_image_pair(scene_name, pair_name,
image2depth, False)
print pair_name
uts.plot_images(OrderedDict([('image', image1),
('depth_gt', depth_gt)]),
layout=[4, 2])
continue
depth_gt_down = uts_3d.down_sample_depth(depth_gt,
method='uniform',
percent=rate,
K=params['intrinsic'])
try:
start_time = time.time()
print "\t upsampling {}".format(pair_name)
depth_up = deep_upsampler.UpSample(depth_gt_down,
[image1, image2])
np.save(outfile, depth_up)
print "\t time: {}".format(time.time() - start_time)
all_time += time.time() - start_time
all_count += 1
except:
print "{} failed".format(pair_name)
if vis:
uts.plot_images(OrderedDict([('image', image1),
('depth_gt', depth_gt),
('depth_up', depth_up)]),
layout=[4, 2])
print "average run time {}\n".format(all_time / all_count)