Spaces:
Running
on
Zero
Running
on
Zero
Update
Browse files
app.py
CHANGED
@@ -2,13 +2,13 @@
|
|
2 |
|
3 |
from __future__ import annotations
|
4 |
|
5 |
-
import os
|
6 |
import pathlib
|
7 |
import sys
|
8 |
|
9 |
import cv2
|
10 |
import gradio as gr
|
11 |
import numpy as np
|
|
|
12 |
import torch
|
13 |
|
14 |
sys.path.insert(0, "face_detection")
|
@@ -20,15 +20,29 @@ from ibug.face_detection import RetinaFacePredictor
|
|
20 |
DESCRIPTION = "# [ibug-group/face_alignment](https://github.com/ibug-group/face_alignment)"
|
21 |
|
22 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
23 |
-
detector = RetinaFacePredictor(threshold=0.8, device=
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
model_names = [
|
25 |
"2dfan2",
|
26 |
"2dfan4",
|
27 |
"2dfan2_alt",
|
28 |
]
|
29 |
-
models = {name:
|
30 |
|
31 |
|
|
|
32 |
def predict(image: np.ndarray, model_name: str, max_num_faces: int, landmark_score_threshold: int) -> np.ndarray:
|
33 |
model = models[model_name]
|
34 |
|
@@ -72,7 +86,6 @@ with gr.Blocks(css="style.css") as demo:
|
|
72 |
inputs=[image, model_name, max_num_faces, landmark_score_thrshold],
|
73 |
outputs=result,
|
74 |
fn=predict,
|
75 |
-
cache_examples=os.getenv("CACHE_EXAMPLES") == "1",
|
76 |
)
|
77 |
run_button.click(
|
78 |
fn=predict,
|
|
|
2 |
|
3 |
from __future__ import annotations
|
4 |
|
|
|
5 |
import pathlib
|
6 |
import sys
|
7 |
|
8 |
import cv2
|
9 |
import gradio as gr
|
10 |
import numpy as np
|
11 |
+
import spaces
|
12 |
import torch
|
13 |
|
14 |
sys.path.insert(0, "face_detection")
|
|
|
20 |
DESCRIPTION = "# [ibug-group/face_alignment](https://github.com/ibug-group/face_alignment)"
|
21 |
|
22 |
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
|
23 |
+
detector = RetinaFacePredictor(threshold=0.8, device="cpu", model=RetinaFacePredictor.get_model("mobilenet0.25"))
|
24 |
+
detector.device = device
|
25 |
+
detector.net.to(device)
|
26 |
+
|
27 |
+
|
28 |
+
def load_model(model_name: str, device: torch.device) -> FANPredictor:
|
29 |
+
model = FANPredictor(
|
30 |
+
device="cpu", model=FANPredictor.get_model(model_name), config=FANPredictor.create_config(use_jit=False)
|
31 |
+
)
|
32 |
+
model.device = device
|
33 |
+
model.net.to(device)
|
34 |
+
return model
|
35 |
+
|
36 |
+
|
37 |
model_names = [
|
38 |
"2dfan2",
|
39 |
"2dfan4",
|
40 |
"2dfan2_alt",
|
41 |
]
|
42 |
+
models = {name: load_model(name, device) for name in model_names}
|
43 |
|
44 |
|
45 |
+
@spaces.GPU
|
46 |
def predict(image: np.ndarray, model_name: str, max_num_faces: int, landmark_score_threshold: int) -> np.ndarray:
|
47 |
model = models[model_name]
|
48 |
|
|
|
86 |
inputs=[image, model_name, max_num_faces, landmark_score_thrshold],
|
87 |
outputs=result,
|
88 |
fn=predict,
|
|
|
89 |
)
|
90 |
run_button.click(
|
91 |
fn=predict,
|