-
Notifications
You must be signed in to change notification settings - Fork 22
/
Copy pathrenat.yaml
59 lines (53 loc) · 1.97 KB
/
renat.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
data_dirpath: "/home/renat/data/rgbd-kinect-pose"
output_dirpath: "/home/renat/Desktop/kavatar_test"
aggregate: # aggregate results from body, hand and face blocks
log_level: 1
gender: "male" # which gender and shape to use for visualization and filtering
person_shape_path: "shapefit/renat"
vis_pose:
enable: True
imshow: True
imsave: False
scale: .2 # reduce resolution for real-time visualization
device: "cuda:0" # must be cuda: minimal_pytorch_rasterizer for mesh visualizaion does not support cpu
filterer:
device: "cpu" # for smplx inference
modify_wrist: False # True value provides good result only on slow motions with a clear background
modify_wrist_th: 0.05
modify_wrist_N: 10
filter_wrist: True # wrist here is a body joint to which hand is attached
filter_hand: True # filter each of 15 joints for each hand
filter_global_trans: True # global translation for body
filter_global_rot: True # global rotation for body
filter_body_pose: True # filter 21 joints of body
filter_jaw_pose: True # filter single jaw joint
filter_face_expression: True # filter 10 face expression parameters
fix_global: False
k4a: # kinect streaming block
log_level: 2
dump_fp: null
# dump_fp: "/storage/wacv_publish/pyk4a_dump/test.pickle"
skip_old: False
fps: 30
parallel_bt: True
gpu_id: 0
hand_pose:
enable: True
log_level: 1
hand_mesh_model_path: "minimal_hand/model/hand_mesh/hand_mesh_model.pkl"
detection_model_path: "minimal_hand/model/detnet/detnet.ckpt"
ik_model_path: "minimal_hand/model/iknet/iknet.ckpt"
gpu_id: 0
face_pose:
enable: True
log_level: 1
config_path: "face_expression/04/config.yaml"
checkpoint_path: "face_expression/04/checkpoint_000044.pth"
device: "cpu"
body_pose:
log_level: 1
model_path: "smplx_kinect/04"
checkpoint: 110000
gender: "male" # gender and shape of a person that is in front of the camera
person_shape_path: "shapefit/renat"
device: "cuda:0"