import os | |
import os.path as osp | |
import sys | |
class Config: | |
## model setting | |
resnet_type = 50 | |
hand_resnet_type = 50 | |
face_resnet_type = 18 | |
model_type = 'OSX' | |
# osx_model_settiing | |
upscale = 4 | |
with_wrist = False | |
hand_pos_joint_num = 20 | |
num_task_token = 24 | |
feat_dim = 1024 | |
encoder_config_file = 'grounded-sam-osx/transformer_utils/configs/osx/encoder/body_encoder_large.py' | |
## input, output | |
input_img_shape = (512, 384) | |
input_body_shape = (256, 192) | |
output_hm_shape = (16, 16, 12) | |
# output_hm_shape = (8, 8, 6) | |
input_hand_shape = (256, 256) | |
# output_hand_hm_shape = (8, 8, 8) | |
output_hand_hm_shape = (16, 16, 16) | |
input_face_shape = (192, 192) | |
focal = (5000, 5000) # virtual focal lengths | |
princpt = (input_body_shape[1]/2, input_body_shape[0]/2) # virtual principal point position | |
body_3d_size = 2 | |
hand_3d_size = 0.3 | |
face_3d_size = 0.3 | |
camera_3d_size = 2.5 | |
## human models | |
flame_shape_params = 100 | |
flame_expression_params = 50 | |
## directory | |
cur_dir = osp.dirname(os.path.abspath(__file__)) | |
human_model_path = osp.join(cur_dir, 'utils', 'human_model_files') | |
cfg = Config() | |