Spaces:
Running
Running
hysts
commited on
Commit
•
e89b524
1
Parent(s):
f77ad3f
Fix #2
Browse files
app.py
CHANGED
@@ -57,6 +57,7 @@ def main():
|
|
57 |
input_image = gr.Image(label='Input Pose Image',
|
58 |
type='pil',
|
59 |
elem_id='input-image')
|
|
|
60 |
with gr.Row():
|
61 |
paths = sorted(pathlib.Path('pose_images').glob('*.png'))
|
62 |
example_images = gr.Dataset(components=[input_image],
|
@@ -112,18 +113,22 @@ Note: Currently, only 5 types of textures are supported, i.e., pure color, strip
|
|
112 |
gr.Markdown(FOOTER)
|
113 |
|
114 |
input_image.change(fn=model.process_pose_image,
|
115 |
-
inputs=
|
116 |
-
outputs=
|
117 |
generate_label_button.click(fn=model.generate_label_image,
|
118 |
-
inputs=[
|
119 |
-
|
|
|
|
|
|
|
120 |
generate_human_button.click(fn=model.generate_human,
|
121 |
inputs=[
|
|
|
122 |
texture_text,
|
123 |
sample_steps,
|
124 |
seed,
|
125 |
],
|
126 |
-
outputs=
|
127 |
example_images.click(fn=set_example_image,
|
128 |
inputs=example_images,
|
129 |
outputs=example_images.components)
|
|
|
57 |
input_image = gr.Image(label='Input Pose Image',
|
58 |
type='pil',
|
59 |
elem_id='input-image')
|
60 |
+
pose_data = gr.Variable()
|
61 |
with gr.Row():
|
62 |
paths = sorted(pathlib.Path('pose_images').glob('*.png'))
|
63 |
example_images = gr.Dataset(components=[input_image],
|
|
|
113 |
gr.Markdown(FOOTER)
|
114 |
|
115 |
input_image.change(fn=model.process_pose_image,
|
116 |
+
inputs=input_image,
|
117 |
+
outputs=pose_data)
|
118 |
generate_label_button.click(fn=model.generate_label_image,
|
119 |
+
inputs=[
|
120 |
+
pose_data,
|
121 |
+
shape_text,
|
122 |
+
],
|
123 |
+
outputs=label_image)
|
124 |
generate_human_button.click(fn=model.generate_human,
|
125 |
inputs=[
|
126 |
+
label_image,
|
127 |
texture_text,
|
128 |
sample_steps,
|
129 |
seed,
|
130 |
],
|
131 |
+
outputs=result)
|
132 |
example_images.click(fn=set_example_image,
|
133 |
inputs=example_images,
|
134 |
outputs=example_images.components)
|
model.py
CHANGED
@@ -98,29 +98,32 @@ class Model:
|
|
98 |
result = np.asarray(result[0, :, :, :], dtype=np.uint8)
|
99 |
return result
|
100 |
|
101 |
-
def process_pose_image(self, pose_image: PIL.Image.Image) ->
|
102 |
if pose_image is None:
|
103 |
return
|
104 |
data = self.preprocess_pose_image(pose_image)
|
105 |
self.model.feed_pose_data(data)
|
|
|
106 |
|
107 |
-
def generate_label_image(self,
|
|
|
|
|
108 |
shape_attributes = generate_shape_attributes(shape_text)
|
109 |
shape_attributes = torch.LongTensor(shape_attributes).unsqueeze(0)
|
110 |
self.model.feed_shape_attributes(shape_attributes)
|
111 |
self.model.generate_parsing_map()
|
112 |
self.model.generate_quantized_segm()
|
113 |
colored_segm = self.model.palette_result(self.model.segm[0].cpu())
|
|
|
114 |
|
115 |
-
|
|
|
|
|
116 |
seg_map = self.process_mask(mask)
|
117 |
self.model.segm = torch.from_numpy(seg_map).unsqueeze(0).unsqueeze(
|
118 |
0).to(self.model.device)
|
119 |
self.model.generate_quantized_segm()
|
120 |
-
return colored_segm
|
121 |
|
122 |
-
def generate_human(self, texture_text: str, sample_steps: int,
|
123 |
-
seed: int) -> np.ndarray:
|
124 |
set_random_seed(seed)
|
125 |
|
126 |
texture_attributes = generate_texture_attributes(texture_text)
|
|
|
98 |
result = np.asarray(result[0, :, :, :], dtype=np.uint8)
|
99 |
return result
|
100 |
|
101 |
+
def process_pose_image(self, pose_image: PIL.Image.Image) -> torch.Tensor:
|
102 |
if pose_image is None:
|
103 |
return
|
104 |
data = self.preprocess_pose_image(pose_image)
|
105 |
self.model.feed_pose_data(data)
|
106 |
+
return data
|
107 |
|
108 |
+
def generate_label_image(self, pose_data: torch.Tensor,
|
109 |
+
shape_text: str) -> np.ndarray:
|
110 |
+
self.model.feed_pose_data(pose_data)
|
111 |
shape_attributes = generate_shape_attributes(shape_text)
|
112 |
shape_attributes = torch.LongTensor(shape_attributes).unsqueeze(0)
|
113 |
self.model.feed_shape_attributes(shape_attributes)
|
114 |
self.model.generate_parsing_map()
|
115 |
self.model.generate_quantized_segm()
|
116 |
colored_segm = self.model.palette_result(self.model.segm[0].cpu())
|
117 |
+
return colored_segm
|
118 |
|
119 |
+
def generate_human(self, label_image: np.ndarray, texture_text: str,
|
120 |
+
sample_steps: int, seed: int) -> np.ndarray:
|
121 |
+
mask = label_image.copy()
|
122 |
seg_map = self.process_mask(mask)
|
123 |
self.model.segm = torch.from_numpy(seg_map).unsqueeze(0).unsqueeze(
|
124 |
0).to(self.model.device)
|
125 |
self.model.generate_quantized_segm()
|
|
|
126 |
|
|
|
|
|
127 |
set_random_seed(seed)
|
128 |
|
129 |
texture_attributes = generate_texture_attributes(texture_text)
|