Spaces:
Running
on
Zero
Running
on
Zero
zzzweakman
commited on
Commit
·
0bc2c6f
1
Parent(s):
5379bd5
fix: retargeting feature leakage
Browse files- app.py +22 -11
- assets/gradio_description_retargeting.md +1 -1
- src/gradio_pipeline.py +22 -50
app.py
CHANGED
|
@@ -72,7 +72,7 @@ data_examples = [
|
|
| 72 |
# Define components first
|
| 73 |
eye_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target eyes-open ratio")
|
| 74 |
lip_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target lip-open ratio")
|
| 75 |
-
retargeting_input_image = gr.Image(type="
|
| 76 |
output_image = gr.Image(type="numpy")
|
| 77 |
output_image_paste_back = gr.Image(type="numpy")
|
| 78 |
output_video = gr.Video()
|
|
@@ -144,11 +144,11 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 144 |
examples_per_page=5,
|
| 145 |
cache_examples=False,
|
| 146 |
)
|
| 147 |
-
gr.Markdown(load_description("assets/gradio_description_retargeting.md"), visible=
|
| 148 |
-
with gr.Row(visible=
|
| 149 |
eye_retargeting_slider.render()
|
| 150 |
lip_retargeting_slider.render()
|
| 151 |
-
with gr.Row(visible=
|
| 152 |
process_button_retargeting = gr.Button("🚗 Retargeting", variant="primary")
|
| 153 |
process_button_reset_retargeting = gr.ClearButton(
|
| 154 |
[
|
|
@@ -160,10 +160,21 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 160 |
],
|
| 161 |
value="🧹 Clear"
|
| 162 |
)
|
| 163 |
-
with gr.Row(visible=
|
| 164 |
with gr.Column():
|
| 165 |
with gr.Accordion(open=True, label="Retargeting Input"):
|
| 166 |
retargeting_input_image.render()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 167 |
with gr.Column():
|
| 168 |
with gr.Accordion(open=True, label="Retargeting Result"):
|
| 169 |
output_image.render()
|
|
@@ -174,7 +185,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 174 |
process_button_retargeting.click(
|
| 175 |
# fn=gradio_pipeline.execute_image,
|
| 176 |
fn=gpu_wrapped_execute_image,
|
| 177 |
-
inputs=[eye_retargeting_slider, lip_retargeting_slider],
|
| 178 |
outputs=[output_image, output_image_paste_back],
|
| 179 |
show_progress=True
|
| 180 |
)
|
|
@@ -190,11 +201,11 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
| 190 |
outputs=[output_video, output_video_concat],
|
| 191 |
show_progress=True
|
| 192 |
)
|
| 193 |
-
image_input.change(
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
)
|
| 198 |
video_input.upload(
|
| 199 |
fn=is_square_video,
|
| 200 |
inputs=video_input,
|
|
|
|
| 72 |
# Define components first
|
| 73 |
eye_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target eyes-open ratio")
|
| 74 |
lip_retargeting_slider = gr.Slider(minimum=0, maximum=0.8, step=0.01, label="target lip-open ratio")
|
| 75 |
+
retargeting_input_image = gr.Image(type="filepath")
|
| 76 |
output_image = gr.Image(type="numpy")
|
| 77 |
output_image_paste_back = gr.Image(type="numpy")
|
| 78 |
output_video = gr.Video()
|
|
|
|
| 144 |
examples_per_page=5,
|
| 145 |
cache_examples=False,
|
| 146 |
)
|
| 147 |
+
gr.Markdown(load_description("assets/gradio_description_retargeting.md"), visible=True)
|
| 148 |
+
with gr.Row(visible=True):
|
| 149 |
eye_retargeting_slider.render()
|
| 150 |
lip_retargeting_slider.render()
|
| 151 |
+
with gr.Row(visible=True):
|
| 152 |
process_button_retargeting = gr.Button("🚗 Retargeting", variant="primary")
|
| 153 |
process_button_reset_retargeting = gr.ClearButton(
|
| 154 |
[
|
|
|
|
| 160 |
],
|
| 161 |
value="🧹 Clear"
|
| 162 |
)
|
| 163 |
+
with gr.Row(visible=True):
|
| 164 |
with gr.Column():
|
| 165 |
with gr.Accordion(open=True, label="Retargeting Input"):
|
| 166 |
retargeting_input_image.render()
|
| 167 |
+
gr.Examples(
|
| 168 |
+
examples=[
|
| 169 |
+
[osp.join(example_portrait_dir, "s9.jpg")],
|
| 170 |
+
[osp.join(example_portrait_dir, "s6.jpg")],
|
| 171 |
+
[osp.join(example_portrait_dir, "s10.jpg")],
|
| 172 |
+
[osp.join(example_portrait_dir, "s5.jpg")],
|
| 173 |
+
[osp.join(example_portrait_dir, "s7.jpg")],
|
| 174 |
+
],
|
| 175 |
+
inputs=[retargeting_input_image],
|
| 176 |
+
cache_examples=False,
|
| 177 |
+
)
|
| 178 |
with gr.Column():
|
| 179 |
with gr.Accordion(open=True, label="Retargeting Result"):
|
| 180 |
output_image.render()
|
|
|
|
| 185 |
process_button_retargeting.click(
|
| 186 |
# fn=gradio_pipeline.execute_image,
|
| 187 |
fn=gpu_wrapped_execute_image,
|
| 188 |
+
inputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image, flag_do_crop_input],
|
| 189 |
outputs=[output_image, output_image_paste_back],
|
| 190 |
show_progress=True
|
| 191 |
)
|
|
|
|
| 201 |
outputs=[output_video, output_video_concat],
|
| 202 |
show_progress=True
|
| 203 |
)
|
| 204 |
+
# image_input.change(
|
| 205 |
+
# fn=gradio_pipeline.prepare_retargeting,
|
| 206 |
+
# inputs=image_input,
|
| 207 |
+
# outputs=[eye_retargeting_slider, lip_retargeting_slider, retargeting_input_image]
|
| 208 |
+
# )
|
| 209 |
video_input.upload(
|
| 210 |
fn=is_square_video,
|
| 211 |
inputs=video_input,
|
assets/gradio_description_retargeting.md
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
<span style="font-size: 1.2em;">🔥 To change the
|
|
|
|
| 1 |
+
<span style="font-size: 1.2em;">🔥 To change the eyes and lip open ratio of the source portrait, please drag the sliders and then click the <strong>🚗 Retargeting</strong> button. The result would be shown in the blocks. You can try running it multiple times. <strong>😊 Set both ratios to 0.8 to see what's going on!</strong> </span>
|
src/gradio_pipeline.py
CHANGED
|
@@ -26,16 +26,6 @@ class GradioPipeline(LivePortraitPipeline):
|
|
| 26 |
super().__init__(inference_cfg, crop_cfg)
|
| 27 |
# self.live_portrait_wrapper = self.live_portrait_wrapper
|
| 28 |
self.args = args
|
| 29 |
-
# for single image retargeting
|
| 30 |
-
self.start_prepare = False
|
| 31 |
-
self.f_s_user = None
|
| 32 |
-
self.x_c_s_info_user = None
|
| 33 |
-
self.x_s_user = None
|
| 34 |
-
self.source_lmk_user = None
|
| 35 |
-
self.mask_ori = None
|
| 36 |
-
self.img_rgb = None
|
| 37 |
-
self.crop_M_c2o = None
|
| 38 |
-
|
| 39 |
|
| 40 |
def execute_video(
|
| 41 |
self,
|
|
@@ -66,30 +56,23 @@ class GradioPipeline(LivePortraitPipeline):
|
|
| 66 |
else:
|
| 67 |
raise gr.Error("The input source portrait or driving video hasn't been prepared yet 💥!", duration=5)
|
| 68 |
|
| 69 |
-
def execute_image(self, input_eye_ratio: float, input_lip_ratio: float):
|
| 70 |
""" for single image retargeting
|
| 71 |
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
| 72 |
if input_eye_ratio is None or input_eye_ratio is None:
|
| 73 |
raise gr.Error("Invalid ratio input 💥!", duration=5)
|
| 74 |
-
elif self.f_s_user is None:
|
| 75 |
-
if self.start_prepare:
|
| 76 |
-
raise gr.Error(
|
| 77 |
-
"The source portrait is under processing 💥! Please wait for a second.",
|
| 78 |
-
duration=5
|
| 79 |
-
)
|
| 80 |
-
else:
|
| 81 |
-
raise gr.Error(
|
| 82 |
-
"The source portrait hasn't been prepared yet 💥! Please scroll to the top of the page to upload.",
|
| 83 |
-
duration=5
|
| 84 |
-
)
|
| 85 |
else:
|
| 86 |
-
x_s_user =
|
| 87 |
-
f_s_user =
|
| 88 |
# ∆_eyes,i = R_eyes(x_s; c_s,eyes, c_d,eyes,i)
|
| 89 |
-
combined_eye_ratio_tensor = self.live_portrait_wrapper.calc_combined_eye_ratio([[input_eye_ratio]],
|
| 90 |
eyes_delta = self.live_portrait_wrapper.retarget_eye(x_s_user, combined_eye_ratio_tensor)
|
| 91 |
# ∆_lip,i = R_lip(x_s; c_s,lip, c_d,lip,i)
|
| 92 |
-
combined_lip_ratio_tensor = self.live_portrait_wrapper.calc_combined_lip_ratio([[input_lip_ratio]],
|
| 93 |
lip_delta = self.live_portrait_wrapper.retarget_lip(x_s_user, combined_lip_ratio_tensor)
|
| 94 |
num_kp = x_s_user.shape[1]
|
| 95 |
# default: use x_s
|
|
@@ -97,21 +80,20 @@ class GradioPipeline(LivePortraitPipeline):
|
|
| 97 |
# D(W(f_s; x_s, x′_d))
|
| 98 |
out = self.live_portrait_wrapper.warp_decode(f_s_user, x_s_user, x_d_new)
|
| 99 |
out = self.live_portrait_wrapper.parse_output(out['out'])[0]
|
| 100 |
-
out_to_ori_blend = paste_back(out,
|
| 101 |
# gr.Info("Run successfully!", duration=2)
|
| 102 |
return out, out_to_ori_blend
|
| 103 |
|
| 104 |
|
| 105 |
-
def prepare_retargeting(self,
|
| 106 |
""" for single image retargeting
|
| 107 |
"""
|
| 108 |
-
if
|
| 109 |
# gr.Info("Upload successfully!", duration=2)
|
| 110 |
-
self.start_prepare = True
|
| 111 |
inference_cfg = self.live_portrait_wrapper.cfg
|
| 112 |
######## process source portrait ########
|
| 113 |
-
img_rgb = load_img_online(
|
| 114 |
-
log(f"Load source image from {
|
| 115 |
crop_info = self.cropper.crop_single_image(img_rgb)
|
| 116 |
if flag_do_crop:
|
| 117 |
I_s = self.live_portrait_wrapper.prepare_source(crop_info['img_crop_256x256'])
|
|
@@ -120,23 +102,13 @@ class GradioPipeline(LivePortraitPipeline):
|
|
| 120 |
x_s_info = self.live_portrait_wrapper.get_kp_info(I_s)
|
| 121 |
R_s = get_rotation_matrix(x_s_info['pitch'], x_s_info['yaw'], x_s_info['roll'])
|
| 122 |
############################################
|
| 123 |
-
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
self.img_rgb = img_rgb
|
| 130 |
-
self.crop_M_c2o = crop_info['M_c2o']
|
| 131 |
-
self.mask_ori = prepare_paste_back(inference_cfg.mask_crop, crop_info['M_c2o'], dsize=(img_rgb.shape[1], img_rgb.shape[0]))
|
| 132 |
-
# update slider
|
| 133 |
-
eye_close_ratio = calc_eye_close_ratio(self.source_lmk_user[None])
|
| 134 |
-
eye_close_ratio = float(eye_close_ratio.squeeze(0).mean())
|
| 135 |
-
lip_close_ratio = calc_lip_close_ratio(self.source_lmk_user[None])
|
| 136 |
-
lip_close_ratio = float(lip_close_ratio.squeeze(0).mean())
|
| 137 |
-
# for vis
|
| 138 |
-
self.I_s_vis = self.live_portrait_wrapper.parse_output(I_s)[0]
|
| 139 |
-
return eye_close_ratio, lip_close_ratio, self.I_s_vis
|
| 140 |
else:
|
| 141 |
# when press the clear button, go here
|
| 142 |
-
|
|
|
|
|
|
| 26 |
super().__init__(inference_cfg, crop_cfg)
|
| 27 |
# self.live_portrait_wrapper = self.live_portrait_wrapper
|
| 28 |
self.args = args
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
def execute_video(
|
| 31 |
self,
|
|
|
|
| 56 |
else:
|
| 57 |
raise gr.Error("The input source portrait or driving video hasn't been prepared yet 💥!", duration=5)
|
| 58 |
|
| 59 |
+
def execute_image(self, input_eye_ratio: float, input_lip_ratio: float, input_image, flag_do_crop = True):
|
| 60 |
""" for single image retargeting
|
| 61 |
"""
|
| 62 |
+
# disposable feature
|
| 63 |
+
f_s_user, x_s_user, source_lmk_user, crop_M_c2o, mask_ori, img_rgb = \
|
| 64 |
+
self.prepare_retargeting(input_image, flag_do_crop)
|
| 65 |
+
|
| 66 |
if input_eye_ratio is None or input_eye_ratio is None:
|
| 67 |
raise gr.Error("Invalid ratio input 💥!", duration=5)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
else:
|
| 69 |
+
x_s_user = x_s_user.to("cuda")
|
| 70 |
+
f_s_user = f_s_user.to("cuda")
|
| 71 |
# ∆_eyes,i = R_eyes(x_s; c_s,eyes, c_d,eyes,i)
|
| 72 |
+
combined_eye_ratio_tensor = self.live_portrait_wrapper.calc_combined_eye_ratio([[input_eye_ratio]], source_lmk_user)
|
| 73 |
eyes_delta = self.live_portrait_wrapper.retarget_eye(x_s_user, combined_eye_ratio_tensor)
|
| 74 |
# ∆_lip,i = R_lip(x_s; c_s,lip, c_d,lip,i)
|
| 75 |
+
combined_lip_ratio_tensor = self.live_portrait_wrapper.calc_combined_lip_ratio([[input_lip_ratio]], source_lmk_user)
|
| 76 |
lip_delta = self.live_portrait_wrapper.retarget_lip(x_s_user, combined_lip_ratio_tensor)
|
| 77 |
num_kp = x_s_user.shape[1]
|
| 78 |
# default: use x_s
|
|
|
|
| 80 |
# D(W(f_s; x_s, x′_d))
|
| 81 |
out = self.live_portrait_wrapper.warp_decode(f_s_user, x_s_user, x_d_new)
|
| 82 |
out = self.live_portrait_wrapper.parse_output(out['out'])[0]
|
| 83 |
+
out_to_ori_blend = paste_back(out, crop_M_c2o, img_rgb, mask_ori)
|
| 84 |
# gr.Info("Run successfully!", duration=2)
|
| 85 |
return out, out_to_ori_blend
|
| 86 |
|
| 87 |
|
| 88 |
+
def prepare_retargeting(self, input_image, flag_do_crop = True):
|
| 89 |
""" for single image retargeting
|
| 90 |
"""
|
| 91 |
+
if input_image is not None:
|
| 92 |
# gr.Info("Upload successfully!", duration=2)
|
|
|
|
| 93 |
inference_cfg = self.live_portrait_wrapper.cfg
|
| 94 |
######## process source portrait ########
|
| 95 |
+
img_rgb = load_img_online(input_image, mode='rgb', max_dim=1280, n=16)
|
| 96 |
+
log(f"Load source image from {input_image}.")
|
| 97 |
crop_info = self.cropper.crop_single_image(img_rgb)
|
| 98 |
if flag_do_crop:
|
| 99 |
I_s = self.live_portrait_wrapper.prepare_source(crop_info['img_crop_256x256'])
|
|
|
|
| 102 |
x_s_info = self.live_portrait_wrapper.get_kp_info(I_s)
|
| 103 |
R_s = get_rotation_matrix(x_s_info['pitch'], x_s_info['yaw'], x_s_info['roll'])
|
| 104 |
############################################
|
| 105 |
+
f_s_user = self.live_portrait_wrapper.extract_feature_3d(I_s)
|
| 106 |
+
x_s_user = self.live_portrait_wrapper.transform_keypoint(x_s_info)
|
| 107 |
+
source_lmk_user = crop_info['lmk_crop']
|
| 108 |
+
crop_M_c2o = crop_info['M_c2o']
|
| 109 |
+
mask_ori = prepare_paste_back(inference_cfg.mask_crop, crop_info['M_c2o'], dsize=(img_rgb.shape[1], img_rgb.shape[0]))
|
| 110 |
+
return f_s_user, x_s_user, source_lmk_user, crop_M_c2o, mask_ori, img_rgb
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
else:
|
| 112 |
# when press the clear button, go here
|
| 113 |
+
raise gr.Error("The retargeting input hasn't been prepared yet 💥!", duration=5)
|
| 114 |
+
|