lithiumice commited on
Commit
04ade0a
·
1 Parent(s): e7fcdee

add readme

Browse files
Files changed (1) hide show
  1. README.md +95 -0
README.md ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Model Repository Documentation
2
+
3
+ ## Repository Structure Overview
4
+
5
+ The repository is organized into eight main directories, each serving a specific purpose in the pipeline:
6
+
7
+ ### Meta Data (1_meta_data)
8
+ Contains AMASS dataset metadata specifically focused on copycat and occlusion information, essential for motion capture applications.
9
+
10
+ ### MediaPipe Models (2_mediapipe_ckpts)
11
+ Houses MediaPipe's specialized models for facial landmarks and hand tracking, providing fundamental capabilities for human pose estimation.
12
+
13
+ ### 4DHumans Framework (3_4DHumans)
14
+ Incorporates the SMPL (Skinned Multi-Person Linear Model) framework along with training artifacts. The directory includes model parameters, joint regressors, and HMR2 (Human Mesh Recovery) training checkpoints with corresponding configuration files.
15
+
16
+ ### SMPLhub (4_SMPLhub)
17
+ Serves as a comprehensive collection of human body models, including:
18
+ - MANO (hand model) parameters for both left and right hands
19
+ - SMPL models in various formats (NPZ and PKL) for male, female, and neutral body types
20
+ - SMPLH (SMPL with detailed hand articulation)
21
+ - SMPLX (extended SMPL model with face and hand expressions)
22
+
23
+ ### Additional Components
24
+ - S3FD (5_S3FD): Contains face detection model weights
25
+ - SyncNet (6_SyncNet): Includes audio-visual synchronization model
26
+ - SGHM (7_SGHM): Houses ResNet50-based model weights
27
+ - KonIQ (8_koniq): Contains pre-trained weights for image quality assessment
28
+
29
+ ```
30
+ ├── 1_meta_data
31
+ │ └── amass_copycat_occlusion_v3.pkl
32
+ ├── 2_mediapipe_ckpts
33
+ │ ├── face_landmarker.task
34
+ │ └── hand_landmarker.task
35
+ ├── 3_4DHumans
36
+ │ ├── data
37
+ │ │ ├── smpl
38
+ │ │ │ └── SMPL_NEUTRAL.pkl
39
+ │ │ ├── smpl_mean_params.npz
40
+ │ │ └── SMPL_to_J19.pkl
41
+ │ └── logs
42
+ │ └── train
43
+ │ └── multiruns
44
+ │ └── hmr2
45
+ │ └── 0
46
+ │ ├── checkpoints
47
+ │ │ └── epoch=35-step=1000000.ckpt
48
+ │ ├── dataset_config.yaml
49
+ │ └── model_config.yaml
50
+ ├── 4_SMPLhub
51
+ │ ├── 0_misc_files
52
+ │ │ └── J_regressor_coco.npy
53
+ │ ├── MANO
54
+ │ │ └── pkl
55
+ │ │ ├── MANO_LEFT.pkl
56
+ │ │ ├── mano_mean_params.npz
57
+ │ │ └── MANO_RIGHT.pkl
58
+ │ ├── SMPL
59
+ │ │ ├── basicmodel_X_lbs_10_207_0_v1.1.0_pkl
60
+ │ │ │ ├── basicmodel_f_lbs_10_207_0_v1.1.0.pkl
61
+ │ │ │ ├── basicmodel_m_lbs_10_207_0_v1.1.0.pkl
62
+ │ │ │ └── basicmodel_neutral_lbs_10_207_0_v1.1.0.pkl
63
+ │ │ ├── X_model_npz
64
+ │ │ │ ├── SMPL_F_model.npz
65
+ │ │ │ ├── SMPL_M_model.npz
66
+ │ │ │ └── SMPL_N_model.npz
67
+ │ │ └── X_pkl
68
+ │ │ ├── SMPL_FEMALE.pkl
69
+ │ │ ├── SMPL_MALE.pkl
70
+ │ │ └── SMPL_NEUTRAL.pkl
71
+ │ ├── SMPLH
72
+ │ │ ├── X_npz
73
+ │ │ │ ├── SMPLH_FEMALE.npz
74
+ │ │ │ ├── SMPLH_MALE.npz
75
+ │ │ │ └── SMPLH_NEUTRAL.npz
76
+ │ │ └── X_pkl
77
+ │ │ ├── SMPLH_female.pkl
78
+ │ │ ├── SMPLH_male.pkl
79
+ │ │ └── SMPLH_NEUTRAL.pkl
80
+ │ └── SMPLX
81
+ │ ├── mod
82
+ │ │ └── SMPLX_MALE_shape2019_exp2020.npz
83
+ │ └── X_npz
84
+ │ ├── SMPLX_FEMALE.npz
85
+ │ ├── SMPLX_MALE.npz
86
+ │ └── SMPLX_NEUTRAL.npz
87
+ ├── 5_S3FD
88
+ │ └── sfd_face.pth
89
+ ├── 6_SyncNet
90
+ │ └── syncnet_v2.model
91
+ ├── 7_SGHM
92
+ │ └── SGHM-ResNet50.pth
93
+ └── 8_koniq
94
+ └── koniq_pretrained.pkl
95
+ ```