config = tf.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.Session(config=config)
dataset=sys.argv[3]
vis=False
output_dir = cfg['path_to_output']
output_img = output_dir+"/" +dataset
if not(os.path.exists(output_img)):
os.makedirs(output_img)
bop_dir,test_dir,model_plys,\
model_info,model_ids,rgb_files,\
depth_files,mask_files,gts,\
cam_param_global,scene_cam = bop_io.get_dataset(cfg,dataset,incl_param=True,train=False)
im_width,im_height =cam_param_global['im_size']
cam_K = cam_param_global['K']
model_params =inout.load_json(os.path.join(bop_dir+"/models_xyz/",cfg['norm_factor_fn']))
if(dataset=='itodd'):
img_type='gray'
else:
img_type='rgb'
if("target_obj" in cfg.keys()):
target_obj = cfg['target_obj']
remove_obj_id=[]
incl_obj_id=[]
def rmfield(a, *fieldnames_to_remove):
return a[[
name for name in a.dtype.names if name not in fieldnames_to_remove
]]
if (len(sys.argv) < 2):
print(
"python3 tools/2_1_ply_file_to_3d_coord_model.py [cfg_fn] [dataset_name]"
)
cfg_fn = sys.argv[1]
cfg = inout.load_json(cfg_fn)
dataset = sys.argv[2]
bop_dir, source_dir, model_plys, model_info, model_ids, rgb_files, depth_files, mask_files, gts, cam_param_global = bop_io.get_dataset(
cfg, dataset)
if not (os.path.exists(bop_dir + "/models_xyz/")):
os.makedirs(bop_dir + "/models_xyz/")
norm_factor = bop_dir + "/models_xyz/" + "norm_factor.json"
param = {}
for m_id, model_ply in enumerate(model_plys):
model_id = model_ids[m_id]
m_info = model_info['{}'.format(model_id)]
keys = m_info.keys()
sym_continous = [0, 0, 0, 0, 0, 0]
center_x = center_y = center_z = True
if ('symmetries_discrete' in keys):
center_x = center_y = center_z = False
print("keep origins of the object when it has symmetric poses")
if p > 0:
y_disc[:] = 0
return X_disc, y_disc
loss_weights = [100, 1]
train_gen_first = False
load_recent_weight = True
dataset = sys.argv[3]
cfg_fn = sys.argv[2] #"cfg/cfg_bop2019.json"
cfg = inout.load_json(cfg_fn)
bop_dir, source_dir, model_plys, model_info, model_ids, rgb_files, depth_files, mask_files, gts, cam_param_global, scene_cam = bop_io.get_dataset(
cfg, dataset, incl_param=True)
im_width, im_height = cam_param_global['im_size']
weight_prefix = "pix2pose"
obj_id = int(sys.argv[4]) #identical to the number for the ply file.
weight_dir = bop_dir + "/pix2pose_weights/{:02d}".format(obj_id)
if not (os.path.exists(weight_dir)):
os.makedirs(weight_dir)
back_dir = sys.argv[5]
data_dir = bop_dir + "/train_xyz/{:02d}".format(obj_id)
batch_size = 50
datagenerator = dataio.data_generator(data_dir,
back_dir,
batch_size=batch_size,
res_x=im_width,
res_y=im_height)
augment_inplane = 30
if len(sys.argv) < 3:
print(
"rendering 3d coordinate images using a converted ply file, format of 6D pose challange(http://cmp.felk.cvut.cz/sixd/challenge_2017/) can be used"
)
print(
"python3 tools/2_2_render_pix2pose_training.py [cfg_fn] [dataset_name]"
)
else:
cfg_fn = sys.argv[1] #"cfg/cfg_bop2019.json"
cfg = inout.load_json(cfg_fn)
dataset = sys.argv[2]
bop_dir,source_dir,model_plys,model_info,model_ids,rgb_files,\
depth_files,mask_files,gts,cam_param_global,scene_cam =\
bop_io.get_dataset(cfg,dataset,incl_param=True)
xyz_target_dir = bop_dir + "/train_xyz"
im_width, im_height = cam_param_global['im_size']
cam_K = cam_param_global['K']
#check if the image dimension is the same
rgb_fn = rgb_files[0]
img_temp = inout.load_im(rgb_fn)
if (img_temp.shape[0] != im_height or img_temp.shape[1] != im_width):
print("the size of training images is different from test images")
im_height = img_temp.shape[0]
im_width = img_temp.shape[1]
ren = Renderer((im_width, im_height), cam_K)
t_model = -1
if (len(sys.argv) == 3):
from mrcnn import utils
import mrcnn.model as modellib
from mrcnn import visualize
from mrcnn.model import log
from imgaug import augmenters as iaa
from tools.mask_rcnn_util import BopDetectConfig,BopDataset
import skimage
if(len(sys.argv)!=3):
print("python3 tools/1_2_train_maskrcnn.py [cfg_fn] [dataset]")
cfg_fn = sys.argv[1] #"cfg/cfg_bop2019.json"
cfg = inout.load_json(cfg_fn)
dataset=sys.argv[2]
bop_dir,_,_,_,model_ids,_,_,_,_,cam_param_global = bop_io.get_dataset(cfg,dataset,train=True)
im_width,im_height = cam_param_global['im_size']
MODEL_DIR = os.path.join(bop_dir, "weight_detection")
config = BopDetectConfig(dataset=dataset,
num_classes=model_ids.shape[0]+1,#1+len(model_plys),
im_width=im_width,im_height=im_height)
config.display()
dataset_train = BopDataset()
dataset_train.set_dataset(dataset,model_ids,
os.path.join(bop_dir,"train_detect"))
dataset_train.load_dataset()
dataset_train.prepare()
model = modellib.MaskRCNN(mode="training", config=config,
from bop_toolkit_lib import inout
from tools import bop_io
import copy
#YCB(have to check) - > LMO
#HB, ITODD -> T-LESS
ref_gt = inout.load_scene_gt(
os.path.join("/home/kiru/media/hdd/bop/tless/train_render_reconst/000001",
"scene_gt.json"))
ref_camera = inout.load_scene_camera(
os.path.join("/home/kiru/media/hdd/bop/tless/train_render_reconst/000001",
"scene_camera.json"))
#bop_dir,source_dir,model_plys,model_info,model_ids,rgb_files,depth_files,mask_files,gts,cam_param_global = bop_io.get_dataset('hb',train=True)
#bop_dir,source_dir,model_plys,model_info,model_ids,rgb_files,depth_files,mask_files,gts,cam_param_global = bop_io.get_dataset('itodd',train=True)
bop_dir, source_dir, model_plys, model_info, model_ids, rgb_files, depth_files, mask_files, gts, cam_param_global = bop_io.get_dataset(
'ycbv', train=True)
im_width, im_height = cam_param_global['im_size']
camK = cam_param_global['K']
cam_K_list = np.array(camK).reshape(-1)
ren = Renderer((im_width, im_height), camK)
source_dir = bop_dir + "/train"
if not (os.path.exists(source_dir)): os.makedirs(source_dir)
for i in range(len(model_plys)):
target_dir = source_dir + "/{:06d}".format(model_ids[i])
if not (os.path.exists(target_dir)): os.makedirs(target_dir)
if not (os.path.exists(target_dir + "/rgb")):
os.makedirs(target_dir + "/rgb")
if not (os.path.exists(target_dir + "/depth")):