Spaces:
Running
Running
Update
Browse files- .pre-commit-config.yaml +59 -36
- README.md +1 -1
- app.py +74 -77
- model.py +91 -116
.pre-commit-config.yaml
CHANGED
@@ -1,37 +1,60 @@
|
|
1 |
-
exclude: ^(ViTPose/|mmdet_configs/configs/)
|
2 |
repos:
|
3 |
-
- repo: https://github.com/pre-commit/pre-commit-hooks
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
- repo: https://github.com/pre-commit/mirrors-mypy
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
repos:
|
2 |
+
- repo: https://github.com/pre-commit/pre-commit-hooks
|
3 |
+
rev: v4.6.0
|
4 |
+
hooks:
|
5 |
+
- id: check-executables-have-shebangs
|
6 |
+
- id: check-json
|
7 |
+
- id: check-merge-conflict
|
8 |
+
- id: check-shebang-scripts-are-executable
|
9 |
+
- id: check-toml
|
10 |
+
- id: check-yaml
|
11 |
+
- id: end-of-file-fixer
|
12 |
+
- id: mixed-line-ending
|
13 |
+
args: ["--fix=lf"]
|
14 |
+
- id: requirements-txt-fixer
|
15 |
+
- id: trailing-whitespace
|
16 |
+
- repo: https://github.com/myint/docformatter
|
17 |
+
rev: v1.7.5
|
18 |
+
hooks:
|
19 |
+
- id: docformatter
|
20 |
+
args: ["--in-place"]
|
21 |
+
- repo: https://github.com/pycqa/isort
|
22 |
+
rev: 5.13.2
|
23 |
+
hooks:
|
24 |
+
- id: isort
|
25 |
+
args: ["--profile", "black"]
|
26 |
+
- repo: https://github.com/pre-commit/mirrors-mypy
|
27 |
+
rev: v1.10.0
|
28 |
+
hooks:
|
29 |
+
- id: mypy
|
30 |
+
args: ["--ignore-missing-imports"]
|
31 |
+
additional_dependencies:
|
32 |
+
[
|
33 |
+
"types-python-slugify",
|
34 |
+
"types-requests",
|
35 |
+
"types-PyYAML",
|
36 |
+
"types-pytz",
|
37 |
+
]
|
38 |
+
- repo: https://github.com/psf/black
|
39 |
+
rev: 24.4.2
|
40 |
+
hooks:
|
41 |
+
- id: black
|
42 |
+
language_version: python3.10
|
43 |
+
args: ["--line-length", "119"]
|
44 |
+
- repo: https://github.com/kynan/nbstripout
|
45 |
+
rev: 0.7.1
|
46 |
+
hooks:
|
47 |
+
- id: nbstripout
|
48 |
+
args:
|
49 |
+
[
|
50 |
+
"--extra-keys",
|
51 |
+
"metadata.interpreter metadata.kernelspec cell.metadata.pycharm",
|
52 |
+
]
|
53 |
+
- repo: https://github.com/nbQA-dev/nbQA
|
54 |
+
rev: 1.8.5
|
55 |
+
hooks:
|
56 |
+
- id: nbqa-black
|
57 |
+
- id: nbqa-pyupgrade
|
58 |
+
args: ["--py37-plus"]
|
59 |
+
- id: nbqa-isort
|
60 |
+
args: ["--float-to-top"]
|
README.md
CHANGED
@@ -4,7 +4,7 @@ emoji: 🦀
|
|
4 |
colorFrom: gray
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
-
sdk_version:
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
suggested_hardware: t4-small
|
|
|
4 |
colorFrom: gray
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
+
sdk_version: 4.36.1
|
8 |
app_file: app.py
|
9 |
pinned: false
|
10 |
suggested_hardware: t4-small
|
app.py
CHANGED
@@ -2,109 +2,106 @@
|
|
2 |
|
3 |
from __future__ import annotations
|
4 |
|
|
|
5 |
import pathlib
|
|
|
|
|
6 |
import tarfile
|
7 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
import gradio as gr
|
9 |
|
10 |
from model import AppModel
|
11 |
|
12 |
-
DESCRIPTION =
|
13 |
|
14 |
Related app: [https://huggingface.co/spaces/Gradio-Blocks/ViTPose](https://huggingface.co/spaces/Gradio-Blocks/ViTPose)
|
15 |
-
|
16 |
|
17 |
|
18 |
def extract_tar() -> None:
|
19 |
-
if pathlib.Path(
|
20 |
return
|
21 |
-
with tarfile.open(
|
22 |
-
f.extractall(
|
23 |
|
24 |
|
25 |
extract_tar()
|
26 |
|
27 |
model = AppModel()
|
28 |
|
29 |
-
with gr.Blocks(css=
|
30 |
gr.Markdown(DESCRIPTION)
|
31 |
|
32 |
with gr.Row():
|
33 |
with gr.Column():
|
34 |
-
input_video = gr.Video(label=
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
choices=list(
|
39 |
-
model.det_model.MODEL_DICT.keys()),
|
40 |
-
value=model.det_model.model_name)
|
41 |
pose_model_name = gr.Dropdown(
|
42 |
-
label=
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
minimum=1,
|
52 |
-
maximum=300,
|
53 |
-
step=1,
|
54 |
-
value=60)
|
55 |
-
predict_button = gr.Button('Predict')
|
56 |
-
pose_preds = gr.Variable()
|
57 |
-
|
58 |
-
paths = sorted(pathlib.Path('videos').rglob('*.mp4'))
|
59 |
-
gr.Examples(examples=[[path.as_posix()] for path in paths],
|
60 |
-
inputs=input_video)
|
61 |
|
62 |
with gr.Column():
|
63 |
-
result = gr.Video(label=
|
64 |
vis_kpt_score_threshold = gr.Slider(
|
65 |
-
label=
|
66 |
-
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
-
vis_dot_radius = gr.Slider(label='Dot Radius',
|
71 |
-
minimum=1,
|
72 |
-
maximum=10,
|
73 |
-
step=1,
|
74 |
-
value=4)
|
75 |
-
vis_line_thickness = gr.Slider(label='Line Thickness',
|
76 |
-
minimum=1,
|
77 |
-
maximum=10,
|
78 |
-
step=1,
|
79 |
-
value=2)
|
80 |
-
redraw_button = gr.Button('Redraw')
|
81 |
|
82 |
detector_name.change(fn=model.det_model.set_model, inputs=detector_name)
|
83 |
-
pose_model_name.change(fn=model.pose_model.set_model,
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
|
93 |
-
|
94 |
-
|
95 |
-
|
96 |
-
|
97 |
-
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
from __future__ import annotations
|
4 |
|
5 |
+
import os
|
6 |
import pathlib
|
7 |
+
import shlex
|
8 |
+
import subprocess
|
9 |
import tarfile
|
10 |
|
11 |
+
if os.getenv("SYSTEM") == "spaces":
|
12 |
+
subprocess.run(shlex.split("pip install click==7.1.2"))
|
13 |
+
subprocess.run(shlex.split("pip install typer==0.9.4"))
|
14 |
+
|
15 |
+
import mim
|
16 |
+
|
17 |
+
mim.uninstall("mmcv-full", confirm_yes=True)
|
18 |
+
mim.install("mmcv-full==1.5.0", is_yes=True)
|
19 |
+
|
20 |
+
subprocess.call(shlex.split("pip uninstall -y opencv-python"))
|
21 |
+
subprocess.call(shlex.split("pip uninstall -y opencv-python-headless"))
|
22 |
+
subprocess.call(shlex.split("pip install opencv-python-headless==4.8.0.74"))
|
23 |
+
|
24 |
+
|
25 |
import gradio as gr
|
26 |
|
27 |
from model import AppModel
|
28 |
|
29 |
+
DESCRIPTION = """# [ViTPose](https://github.com/ViTAE-Transformer/ViTPose)
|
30 |
|
31 |
Related app: [https://huggingface.co/spaces/Gradio-Blocks/ViTPose](https://huggingface.co/spaces/Gradio-Blocks/ViTPose)
|
32 |
+
"""
|
33 |
|
34 |
|
35 |
def extract_tar() -> None:
|
36 |
+
if pathlib.Path("mmdet_configs/configs").exists():
|
37 |
return
|
38 |
+
with tarfile.open("mmdet_configs/configs.tar") as f:
|
39 |
+
f.extractall("mmdet_configs")
|
40 |
|
41 |
|
42 |
extract_tar()
|
43 |
|
44 |
model = AppModel()
|
45 |
|
46 |
+
with gr.Blocks(css="style.css") as demo:
|
47 |
gr.Markdown(DESCRIPTION)
|
48 |
|
49 |
with gr.Row():
|
50 |
with gr.Column():
|
51 |
+
input_video = gr.Video(label="Input Video", format="mp4", elem_id="input_video")
|
52 |
+
detector_name = gr.Dropdown(
|
53 |
+
label="Detector", choices=list(model.det_model.MODEL_DICT.keys()), value=model.det_model.model_name
|
54 |
+
)
|
|
|
|
|
|
|
55 |
pose_model_name = gr.Dropdown(
|
56 |
+
label="Pose Model", choices=list(model.pose_model.MODEL_DICT.keys()), value=model.pose_model.model_name
|
57 |
+
)
|
58 |
+
det_score_threshold = gr.Slider(label="Box Score Threshold", minimum=0, maximum=1, step=0.05, value=0.5)
|
59 |
+
max_num_frames = gr.Slider(label="Maximum Number of Frames", minimum=1, maximum=300, step=1, value=60)
|
60 |
+
predict_button = gr.Button("Predict")
|
61 |
+
pose_preds = gr.State()
|
62 |
+
|
63 |
+
paths = sorted(pathlib.Path("videos").rglob("*.mp4"))
|
64 |
+
gr.Examples(examples=[[path.as_posix()] for path in paths], inputs=input_video)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
|
66 |
with gr.Column():
|
67 |
+
result = gr.Video(label="Result", format="mp4", elem_id="result")
|
68 |
vis_kpt_score_threshold = gr.Slider(
|
69 |
+
label="Visualization Score Threshold", minimum=0, maximum=1, step=0.05, value=0.3
|
70 |
+
)
|
71 |
+
vis_dot_radius = gr.Slider(label="Dot Radius", minimum=1, maximum=10, step=1, value=4)
|
72 |
+
vis_line_thickness = gr.Slider(label="Line Thickness", minimum=1, maximum=10, step=1, value=2)
|
73 |
+
redraw_button = gr.Button("Redraw")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
74 |
|
75 |
detector_name.change(fn=model.det_model.set_model, inputs=detector_name)
|
76 |
+
pose_model_name.change(fn=model.pose_model.set_model, inputs=pose_model_name)
|
77 |
+
predict_button.click(
|
78 |
+
fn=model.run,
|
79 |
+
inputs=[
|
80 |
+
input_video,
|
81 |
+
detector_name,
|
82 |
+
pose_model_name,
|
83 |
+
det_score_threshold,
|
84 |
+
max_num_frames,
|
85 |
+
vis_kpt_score_threshold,
|
86 |
+
vis_dot_radius,
|
87 |
+
vis_line_thickness,
|
88 |
+
],
|
89 |
+
outputs=[
|
90 |
+
result,
|
91 |
+
pose_preds,
|
92 |
+
],
|
93 |
+
)
|
94 |
+
redraw_button.click(
|
95 |
+
fn=model.visualize_pose_results,
|
96 |
+
inputs=[
|
97 |
+
input_video,
|
98 |
+
pose_preds,
|
99 |
+
vis_kpt_score_threshold,
|
100 |
+
vis_dot_radius,
|
101 |
+
vis_line_thickness,
|
102 |
+
],
|
103 |
+
outputs=result,
|
104 |
+
)
|
105 |
+
|
106 |
+
if __name__ == "__main__":
|
107 |
+
demo.queue(max_size=10).launch()
|
model.py
CHANGED
@@ -1,68 +1,49 @@
|
|
1 |
from __future__ import annotations
|
2 |
|
3 |
-
import os
|
4 |
-
import shlex
|
5 |
-
import subprocess
|
6 |
import sys
|
7 |
import tempfile
|
8 |
|
9 |
-
if os.getenv('SYSTEM') == 'spaces':
|
10 |
-
import mim
|
11 |
-
|
12 |
-
mim.uninstall('mmcv-full', confirm_yes=True)
|
13 |
-
mim.install('mmcv-full==1.5.0', is_yes=True)
|
14 |
-
|
15 |
-
subprocess.call(shlex.split('pip uninstall -y opencv-python'))
|
16 |
-
subprocess.call(shlex.split('pip uninstall -y opencv-python-headless'))
|
17 |
-
subprocess.call(
|
18 |
-
shlex.split('pip install opencv-python-headless==4.8.0.74'))
|
19 |
-
|
20 |
import cv2
|
21 |
import huggingface_hub
|
22 |
import numpy as np
|
23 |
import torch
|
24 |
import torch.nn as nn
|
25 |
|
26 |
-
sys.path.insert(0,
|
27 |
|
28 |
from mmdet.apis import inference_detector, init_detector
|
29 |
-
from mmpose.apis import (
|
30 |
-
|
|
|
|
|
|
|
|
|
31 |
|
32 |
|
33 |
class DetModel:
|
34 |
MODEL_DICT = {
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
'model':
|
39 |
-
'https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_tiny_8x8_300e_coco/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth',
|
40 |
},
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
'model':
|
45 |
-
'https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_s_8x8_300e_coco/yolox_s_8x8_300e_coco_20211121_095711-4592a793.pth',
|
46 |
},
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
'model':
|
51 |
-
'https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_l_8x8_300e_coco/yolox_l_8x8_300e_coco_20211126_140236-d3bd2b23.pth',
|
52 |
},
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
'model':
|
57 |
-
'https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_x_8x8_300e_coco/yolox_x_8x8_300e_coco_20211126_140254-1ef88d67.pth',
|
58 |
},
|
59 |
}
|
60 |
|
61 |
def __init__(self):
|
62 |
-
self.device = torch.device(
|
63 |
-
'cuda:0' if torch.cuda.is_available() else 'cpu')
|
64 |
self._load_all_models_once()
|
65 |
-
self.model_name =
|
66 |
self.model = self._load_model(self.model_name)
|
67 |
|
68 |
def _load_all_models_once(self) -> None:
|
@@ -71,7 +52,7 @@ class DetModel:
|
|
71 |
|
72 |
def _load_model(self, name: str) -> nn.Module:
|
73 |
d = self.MODEL_DICT[name]
|
74 |
-
return init_detector(d[
|
75 |
|
76 |
def set_model(self, name: str) -> None:
|
77 |
if name == self.model_name:
|
@@ -79,9 +60,7 @@ class DetModel:
|
|
79 |
self.model_name = name
|
80 |
self.model = self._load_model(name)
|
81 |
|
82 |
-
def detect_and_visualize(
|
83 |
-
self, image: np.ndarray,
|
84 |
-
score_threshold: float) -> tuple[list[np.ndarray], np.ndarray]:
|
85 |
out = self.detect(image)
|
86 |
vis = self.visualize_detection_results(image, out, score_threshold)
|
87 |
return out, vis
|
@@ -92,50 +71,40 @@ class DetModel:
|
|
92 |
return out
|
93 |
|
94 |
def visualize_detection_results(
|
95 |
-
|
96 |
-
|
97 |
-
detection_results: list[np.ndarray],
|
98 |
-
score_threshold: float = 0.3) -> np.ndarray:
|
99 |
person_det = [detection_results[0]] + [np.array([]).reshape(0, 5)] * 79
|
100 |
|
101 |
image = image[:, :, ::-1] # RGB -> BGR
|
102 |
-
vis = self.model.show_result(
|
103 |
-
|
104 |
-
|
105 |
-
bbox_color=None,
|
106 |
-
text_color=(200, 200, 200),
|
107 |
-
mask_color=None)
|
108 |
return vis[:, :, ::-1] # BGR -> RGB
|
109 |
|
110 |
|
111 |
class PoseModel:
|
112 |
MODEL_DICT = {
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
'model': 'models/vitpose-b.pth',
|
117 |
},
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
'model': 'models/vitpose-l.pth',
|
122 |
},
|
123 |
-
|
124 |
-
|
125 |
-
|
126 |
-
'model': 'models/vitpose-b-multi-coco.pth',
|
127 |
},
|
128 |
-
|
129 |
-
|
130 |
-
|
131 |
-
'model': 'models/vitpose-l-multi-coco.pth',
|
132 |
},
|
133 |
}
|
134 |
|
135 |
def __init__(self):
|
136 |
-
self.device = torch.device(
|
137 |
-
|
138 |
-
self.model_name = 'ViTPose-B (multi-task train, COCO)'
|
139 |
self.model = self._load_model(self.model_name)
|
140 |
|
141 |
def _load_all_models_once(self) -> None:
|
@@ -144,9 +113,8 @@ class PoseModel:
|
|
144 |
|
145 |
def _load_model(self, name: str) -> nn.Module:
|
146 |
d = self.MODEL_DICT[name]
|
147 |
-
ckpt_path = huggingface_hub.hf_hub_download(
|
148 |
-
|
149 |
-
model = init_pose_model(d['config'], ckpt_path, device=self.device)
|
150 |
return model
|
151 |
|
152 |
def set_model(self, name: str) -> None:
|
@@ -165,37 +133,36 @@ class PoseModel:
|
|
165 |
vis_line_thickness: int,
|
166 |
) -> tuple[list[dict[str, np.ndarray]], np.ndarray]:
|
167 |
out = self.predict_pose(image, det_results, box_score_threshold)
|
168 |
-
vis = self.visualize_pose_results(image, out, kpt_score_threshold,
|
169 |
-
vis_dot_radius, vis_line_thickness)
|
170 |
return out, vis
|
171 |
|
172 |
def predict_pose(
|
173 |
-
|
174 |
-
|
175 |
-
det_results: list[np.ndarray],
|
176 |
-
box_score_threshold: float = 0.5) -> list[dict[str, np.ndarray]]:
|
177 |
image = image[:, :, ::-1] # RGB -> BGR
|
178 |
person_results = process_mmdet_results(det_results, 1)
|
179 |
-
out, _ = inference_top_down_pose_model(
|
180 |
-
|
181 |
-
|
182 |
-
bbox_thr=box_score_threshold,
|
183 |
-
format='xyxy')
|
184 |
return out
|
185 |
|
186 |
-
def visualize_pose_results(
|
187 |
-
|
188 |
-
|
189 |
-
|
190 |
-
|
191 |
-
|
|
|
|
|
192 |
image = image[:, :, ::-1] # RGB -> BGR
|
193 |
-
vis = vis_pose_result(
|
194 |
-
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
|
|
|
|
|
199 |
return vis[:, :, ::-1] # BGR -> RGB
|
200 |
|
201 |
|
@@ -205,10 +172,15 @@ class AppModel:
|
|
205 |
self.pose_model = PoseModel()
|
206 |
|
207 |
def run(
|
208 |
-
self,
|
209 |
-
|
210 |
-
|
211 |
-
|
|
|
|
|
|
|
|
|
|
|
212 |
) -> tuple[str, list[list[dict[str, np.ndarray]]]]:
|
213 |
if video_path is None:
|
214 |
return
|
@@ -222,8 +194,8 @@ class AppModel:
|
|
222 |
|
223 |
preds_all = []
|
224 |
|
225 |
-
fourcc = cv2.VideoWriter_fourcc(*
|
226 |
-
out_file = tempfile.NamedTemporaryFile(suffix=
|
227 |
writer = cv2.VideoWriter(out_file.name, fourcc, fps, (width, height))
|
228 |
for _ in range(max_num_frames):
|
229 |
ok, frame = cap.read()
|
@@ -232,8 +204,8 @@ class AppModel:
|
|
232 |
rgb_frame = frame[:, :, ::-1]
|
233 |
det_preds = self.det_model.detect(rgb_frame)
|
234 |
preds, vis = self.pose_model.predict_pose_and_visualize(
|
235 |
-
rgb_frame, det_preds, box_score_threshold, kpt_score_threshold,
|
236 |
-
|
237 |
preds_all.append(preds)
|
238 |
writer.write(vis[:, :, ::-1])
|
239 |
cap.release()
|
@@ -241,11 +213,14 @@ class AppModel:
|
|
241 |
|
242 |
return out_file.name, preds_all
|
243 |
|
244 |
-
def visualize_pose_results(
|
245 |
-
|
246 |
-
|
247 |
-
|
248 |
-
|
|
|
|
|
|
|
249 |
if video_path is None or pose_preds_all is None:
|
250 |
return
|
251 |
cap = cv2.VideoCapture(video_path)
|
@@ -253,8 +228,8 @@ class AppModel:
|
|
253 |
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
254 |
fps = cap.get(cv2.CAP_PROP_FPS)
|
255 |
|
256 |
-
fourcc = cv2.VideoWriter_fourcc(*
|
257 |
-
out_file = tempfile.NamedTemporaryFile(suffix=
|
258 |
writer = cv2.VideoWriter(out_file.name, fourcc, fps, (width, height))
|
259 |
for pose_preds in pose_preds_all:
|
260 |
ok, frame = cap.read()
|
@@ -262,8 +237,8 @@ class AppModel:
|
|
262 |
break
|
263 |
rgb_frame = frame[:, :, ::-1]
|
264 |
vis = self.pose_model.visualize_pose_results(
|
265 |
-
rgb_frame, pose_preds, kpt_score_threshold, vis_dot_radius,
|
266 |
-
|
267 |
writer.write(vis[:, :, ::-1])
|
268 |
cap.release()
|
269 |
writer.release()
|
|
|
1 |
from __future__ import annotations
|
2 |
|
|
|
|
|
|
|
3 |
import sys
|
4 |
import tempfile
|
5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
import cv2
|
7 |
import huggingface_hub
|
8 |
import numpy as np
|
9 |
import torch
|
10 |
import torch.nn as nn
|
11 |
|
12 |
+
sys.path.insert(0, "ViTPose/")
|
13 |
|
14 |
from mmdet.apis import inference_detector, init_detector
|
15 |
+
from mmpose.apis import (
|
16 |
+
inference_top_down_pose_model,
|
17 |
+
init_pose_model,
|
18 |
+
process_mmdet_results,
|
19 |
+
vis_pose_result,
|
20 |
+
)
|
21 |
|
22 |
|
23 |
class DetModel:
|
24 |
MODEL_DICT = {
|
25 |
+
"YOLOX-tiny": {
|
26 |
+
"config": "mmdet_configs/configs/yolox/yolox_tiny_8x8_300e_coco.py",
|
27 |
+
"model": "https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_tiny_8x8_300e_coco/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth",
|
|
|
|
|
28 |
},
|
29 |
+
"YOLOX-s": {
|
30 |
+
"config": "mmdet_configs/configs/yolox/yolox_s_8x8_300e_coco.py",
|
31 |
+
"model": "https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_s_8x8_300e_coco/yolox_s_8x8_300e_coco_20211121_095711-4592a793.pth",
|
|
|
|
|
32 |
},
|
33 |
+
"YOLOX-l": {
|
34 |
+
"config": "mmdet_configs/configs/yolox/yolox_l_8x8_300e_coco.py",
|
35 |
+
"model": "https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_l_8x8_300e_coco/yolox_l_8x8_300e_coco_20211126_140236-d3bd2b23.pth",
|
|
|
|
|
36 |
},
|
37 |
+
"YOLOX-x": {
|
38 |
+
"config": "mmdet_configs/configs/yolox/yolox_x_8x8_300e_coco.py",
|
39 |
+
"model": "https://download.openmmlab.com/mmdetection/v2.0/yolox/yolox_x_8x8_300e_coco/yolox_x_8x8_300e_coco_20211126_140254-1ef88d67.pth",
|
|
|
|
|
40 |
},
|
41 |
}
|
42 |
|
43 |
def __init__(self):
|
44 |
+
self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
|
|
45 |
self._load_all_models_once()
|
46 |
+
self.model_name = "YOLOX-l"
|
47 |
self.model = self._load_model(self.model_name)
|
48 |
|
49 |
def _load_all_models_once(self) -> None:
|
|
|
52 |
|
53 |
def _load_model(self, name: str) -> nn.Module:
|
54 |
d = self.MODEL_DICT[name]
|
55 |
+
return init_detector(d["config"], d["model"], device=self.device)
|
56 |
|
57 |
def set_model(self, name: str) -> None:
|
58 |
if name == self.model_name:
|
|
|
60 |
self.model_name = name
|
61 |
self.model = self._load_model(name)
|
62 |
|
63 |
+
def detect_and_visualize(self, image: np.ndarray, score_threshold: float) -> tuple[list[np.ndarray], np.ndarray]:
|
|
|
|
|
64 |
out = self.detect(image)
|
65 |
vis = self.visualize_detection_results(image, out, score_threshold)
|
66 |
return out, vis
|
|
|
71 |
return out
|
72 |
|
73 |
def visualize_detection_results(
|
74 |
+
self, image: np.ndarray, detection_results: list[np.ndarray], score_threshold: float = 0.3
|
75 |
+
) -> np.ndarray:
|
|
|
|
|
76 |
person_det = [detection_results[0]] + [np.array([]).reshape(0, 5)] * 79
|
77 |
|
78 |
image = image[:, :, ::-1] # RGB -> BGR
|
79 |
+
vis = self.model.show_result(
|
80 |
+
image, person_det, score_thr=score_threshold, bbox_color=None, text_color=(200, 200, 200), mask_color=None
|
81 |
+
)
|
|
|
|
|
|
|
82 |
return vis[:, :, ::-1] # BGR -> RGB
|
83 |
|
84 |
|
85 |
class PoseModel:
|
86 |
MODEL_DICT = {
|
87 |
+
"ViTPose-B (single-task train)": {
|
88 |
+
"config": "ViTPose/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py",
|
89 |
+
"model": "models/vitpose-b.pth",
|
|
|
90 |
},
|
91 |
+
"ViTPose-L (single-task train)": {
|
92 |
+
"config": "ViTPose/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py",
|
93 |
+
"model": "models/vitpose-l.pth",
|
|
|
94 |
},
|
95 |
+
"ViTPose-B (multi-task train, COCO)": {
|
96 |
+
"config": "ViTPose/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py",
|
97 |
+
"model": "models/vitpose-b-multi-coco.pth",
|
|
|
98 |
},
|
99 |
+
"ViTPose-L (multi-task train, COCO)": {
|
100 |
+
"config": "ViTPose/configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_large_coco_256x192.py",
|
101 |
+
"model": "models/vitpose-l-multi-coco.pth",
|
|
|
102 |
},
|
103 |
}
|
104 |
|
105 |
def __init__(self):
|
106 |
+
self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
107 |
+
self.model_name = "ViTPose-B (multi-task train, COCO)"
|
|
|
108 |
self.model = self._load_model(self.model_name)
|
109 |
|
110 |
def _load_all_models_once(self) -> None:
|
|
|
113 |
|
114 |
def _load_model(self, name: str) -> nn.Module:
|
115 |
d = self.MODEL_DICT[name]
|
116 |
+
ckpt_path = huggingface_hub.hf_hub_download("public-data/ViTPose", d["model"])
|
117 |
+
model = init_pose_model(d["config"], ckpt_path, device=self.device)
|
|
|
118 |
return model
|
119 |
|
120 |
def set_model(self, name: str) -> None:
|
|
|
133 |
vis_line_thickness: int,
|
134 |
) -> tuple[list[dict[str, np.ndarray]], np.ndarray]:
|
135 |
out = self.predict_pose(image, det_results, box_score_threshold)
|
136 |
+
vis = self.visualize_pose_results(image, out, kpt_score_threshold, vis_dot_radius, vis_line_thickness)
|
|
|
137 |
return out, vis
|
138 |
|
139 |
def predict_pose(
|
140 |
+
self, image: np.ndarray, det_results: list[np.ndarray], box_score_threshold: float = 0.5
|
141 |
+
) -> list[dict[str, np.ndarray]]:
|
|
|
|
|
142 |
image = image[:, :, ::-1] # RGB -> BGR
|
143 |
person_results = process_mmdet_results(det_results, 1)
|
144 |
+
out, _ = inference_top_down_pose_model(
|
145 |
+
self.model, image, person_results=person_results, bbox_thr=box_score_threshold, format="xyxy"
|
146 |
+
)
|
|
|
|
|
147 |
return out
|
148 |
|
149 |
+
def visualize_pose_results(
|
150 |
+
self,
|
151 |
+
image: np.ndarray,
|
152 |
+
pose_results: list[dict[str, np.ndarray]],
|
153 |
+
kpt_score_threshold: float = 0.3,
|
154 |
+
vis_dot_radius: int = 4,
|
155 |
+
vis_line_thickness: int = 1,
|
156 |
+
) -> np.ndarray:
|
157 |
image = image[:, :, ::-1] # RGB -> BGR
|
158 |
+
vis = vis_pose_result(
|
159 |
+
self.model,
|
160 |
+
image,
|
161 |
+
pose_results,
|
162 |
+
kpt_score_thr=kpt_score_threshold,
|
163 |
+
radius=vis_dot_radius,
|
164 |
+
thickness=vis_line_thickness,
|
165 |
+
)
|
166 |
return vis[:, :, ::-1] # BGR -> RGB
|
167 |
|
168 |
|
|
|
172 |
self.pose_model = PoseModel()
|
173 |
|
174 |
def run(
|
175 |
+
self,
|
176 |
+
video_path: str,
|
177 |
+
det_model_name: str,
|
178 |
+
pose_model_name: str,
|
179 |
+
box_score_threshold: float,
|
180 |
+
max_num_frames: int,
|
181 |
+
kpt_score_threshold: float,
|
182 |
+
vis_dot_radius: int,
|
183 |
+
vis_line_thickness: int,
|
184 |
) -> tuple[str, list[list[dict[str, np.ndarray]]]]:
|
185 |
if video_path is None:
|
186 |
return
|
|
|
194 |
|
195 |
preds_all = []
|
196 |
|
197 |
+
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
|
198 |
+
out_file = tempfile.NamedTemporaryFile(suffix=".mp4", delete=False)
|
199 |
writer = cv2.VideoWriter(out_file.name, fourcc, fps, (width, height))
|
200 |
for _ in range(max_num_frames):
|
201 |
ok, frame = cap.read()
|
|
|
204 |
rgb_frame = frame[:, :, ::-1]
|
205 |
det_preds = self.det_model.detect(rgb_frame)
|
206 |
preds, vis = self.pose_model.predict_pose_and_visualize(
|
207 |
+
rgb_frame, det_preds, box_score_threshold, kpt_score_threshold, vis_dot_radius, vis_line_thickness
|
208 |
+
)
|
209 |
preds_all.append(preds)
|
210 |
writer.write(vis[:, :, ::-1])
|
211 |
cap.release()
|
|
|
213 |
|
214 |
return out_file.name, preds_all
|
215 |
|
216 |
+
def visualize_pose_results(
|
217 |
+
self,
|
218 |
+
video_path: str,
|
219 |
+
pose_preds_all: list[list[dict[str, np.ndarray]]],
|
220 |
+
kpt_score_threshold: float,
|
221 |
+
vis_dot_radius: int,
|
222 |
+
vis_line_thickness: int,
|
223 |
+
) -> str:
|
224 |
if video_path is None or pose_preds_all is None:
|
225 |
return
|
226 |
cap = cv2.VideoCapture(video_path)
|
|
|
228 |
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
|
229 |
fps = cap.get(cv2.CAP_PROP_FPS)
|
230 |
|
231 |
+
fourcc = cv2.VideoWriter_fourcc(*"mp4v")
|
232 |
+
out_file = tempfile.NamedTemporaryFile(suffix=".mp4", delete=False)
|
233 |
writer = cv2.VideoWriter(out_file.name, fourcc, fps, (width, height))
|
234 |
for pose_preds in pose_preds_all:
|
235 |
ok, frame = cap.read()
|
|
|
237 |
break
|
238 |
rgb_frame = frame[:, :, ::-1]
|
239 |
vis = self.pose_model.visualize_pose_results(
|
240 |
+
rgb_frame, pose_preds, kpt_score_threshold, vis_dot_radius, vis_line_thickness
|
241 |
+
)
|
242 |
writer.write(vis[:, :, ::-1])
|
243 |
cap.release()
|
244 |
writer.release()
|