Spaces:
Runtime error
Runtime error
Update app.py
Browse filesadded radio i think
app.py
CHANGED
@@ -86,10 +86,9 @@ def generate_annotation(img, overlap=False, hand_encoding=False):
|
|
86 |
annotated_image = draw_landmarks_on_image(image.numpy_view(), detection_result, overlap=overlap, hand_encoding=hand_encoding)
|
87 |
return annotated_image
|
88 |
|
89 |
-
|
90 |
-
|
91 |
-
|
92 |
-
args = Namespace(
|
93 |
pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5",
|
94 |
revision="non-ema",
|
95 |
from_pt=True,
|
@@ -97,8 +96,7 @@ if model_type=="Standard":
|
|
97 |
controlnet_revision=None,
|
98 |
controlnet_from_pt=False,
|
99 |
)
|
100 |
-
|
101 |
-
args = Namespace(
|
102 |
pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5",
|
103 |
revision="non-ema",
|
104 |
from_pt=True,
|
@@ -107,35 +105,58 @@ if model_type=="Hand Encoding":
|
|
107 |
controlnet_from_pt=False,
|
108 |
)
|
109 |
|
110 |
-
|
111 |
-
|
112 |
-
revision=
|
113 |
-
from_pt=
|
|
|
|
|
|
|
|
|
|
|
|
|
114 |
dtype=jnp.float32, # jnp.bfloat16
|
115 |
)
|
116 |
|
117 |
-
|
118 |
-
|
|
|
|
|
119 |
# tokenizer=tokenizer,
|
120 |
-
controlnet=
|
121 |
safety_checker=None,
|
122 |
dtype=jnp.float32, # jnp.bfloat16
|
123 |
-
revision=
|
124 |
-
from_pt=
|
125 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
|
|
|
|
|
127 |
|
128 |
-
|
129 |
-
|
130 |
|
131 |
rng = jax.random.PRNGKey(0)
|
132 |
num_samples = jax.device_count()
|
133 |
prng_seed = jax.random.split(rng, jax.device_count())
|
134 |
|
135 |
|
136 |
-
def infer(prompt, negative_prompt, image):
|
137 |
prompts = num_samples * [prompt]
|
138 |
-
|
|
|
|
|
|
|
139 |
prompt_ids = shard(prompt_ids)
|
140 |
|
141 |
if model_type=="Standard":
|
@@ -145,21 +166,39 @@ def infer(prompt, negative_prompt, image):
|
|
145 |
annotated_image = generate_annotation(image, overlap=False, hand_encoding=True)
|
146 |
overlap_image = generate_annotation(image, overlap=True, hand_encoding=True)
|
147 |
validation_image = Image.fromarray(annotated_image).convert("RGB")
|
148 |
-
processed_image = pipeline.prepare_image_inputs(num_samples * [validation_image])
|
149 |
-
processed_image = shard(processed_image)
|
150 |
|
151 |
-
|
152 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
153 |
|
154 |
-
|
155 |
-
|
156 |
-
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
|
|
|
|
|
|
|
163 |
|
164 |
|
165 |
images = images.reshape((images.shape[0] * images.shape[1],) + images.shape[-3:])
|
@@ -176,16 +215,15 @@ with gr.Blocks(theme='gradio/soft') as demo:
|
|
176 |
Model1 can be found at [https://huggingface.co/Vincent-luo/controlnet-hands](https://huggingface.co/Vincent-luo/controlnet-hands)
|
177 |
|
178 |
Model2 can be found at [https://huggingface.co/MakiPan/controlnet-encoded-hands-130k/ ](https://huggingface.co/MakiPan/controlnet-encoded-hands-130k/)
|
179 |
-
|
180 |
Dataset1 can be found at [https://huggingface.co/datasets/MakiPan/hagrid250k-blip2](https://huggingface.co/datasets/MakiPan/hagrid250k-blip2)
|
181 |
|
182 |
Dataset2 can be found at [https://huggingface.co/datasets/MakiPan/hagrid-hand-enc-250k](https://huggingface.co/datasets/MakiPan/hagrid-hand-enc-250k)
|
183 |
|
184 |
Preprocessing1 can be found at [https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/normal-preprocessing.py](https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/normal-preprocessing.py)
|
185 |
-
|
186 |
Preprocessing2 can be found at [https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/Hand-encoded-preprocessing.py](https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/Hand-encoded-preprocessing.py)
|
187 |
""")
|
188 |
-
|
|
|
189 |
with gr.Row():
|
190 |
with gr.Column():
|
191 |
prompt_input = gr.Textbox(label="Prompt")
|
@@ -227,13 +265,13 @@ with gr.Blocks(theme='gradio/soft') as demo:
|
|
227 |
"example4.png"
|
228 |
],
|
229 |
],
|
230 |
-
inputs=[prompt_input, negative_prompt, input_image],
|
231 |
outputs=[output_image],
|
232 |
fn=infer,
|
233 |
cache_examples=True,
|
234 |
)
|
235 |
|
236 |
-
inputs = [prompt_input, negative_prompt, input_image]
|
237 |
submit_btn.click(fn=infer, inputs=inputs, outputs=[output_image])
|
238 |
|
239 |
demo.launch()
|
|
|
86 |
annotated_image = draw_landmarks_on_image(image.numpy_view(), detection_result, overlap=overlap, hand_encoding=hand_encoding)
|
87 |
return annotated_image
|
88 |
|
89 |
+
|
90 |
+
|
91 |
+
std_args = Namespace(
|
|
|
92 |
pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5",
|
93 |
revision="non-ema",
|
94 |
from_pt=True,
|
|
|
96 |
controlnet_revision=None,
|
97 |
controlnet_from_pt=False,
|
98 |
)
|
99 |
+
enc_args = Namespace(
|
|
|
100 |
pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5",
|
101 |
revision="non-ema",
|
102 |
from_pt=True,
|
|
|
105 |
controlnet_from_pt=False,
|
106 |
)
|
107 |
|
108 |
+
std_controlnet, std_controlnet_params = FlaxControlNetModel.from_pretrained(
|
109 |
+
std_args.controlnet_model_name_or_path,
|
110 |
+
revision=std_args.controlnet_revision,
|
111 |
+
from_pt=std_args.controlnet_from_pt,
|
112 |
+
dtype=jnp.float32, # jnp.bfloat16
|
113 |
+
)
|
114 |
+
enc_controlnet, enc_controlnet_params = FlaxControlNetModel.from_pretrained(
|
115 |
+
enc_args.controlnet_model_name_or_path,
|
116 |
+
revision=enc_args.controlnet_revision,
|
117 |
+
from_pt=enc_args.controlnet_from_pt,
|
118 |
dtype=jnp.float32, # jnp.bfloat16
|
119 |
)
|
120 |
|
121 |
+
|
122 |
+
|
123 |
+
std_pipeline, std_pipeline_params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
|
124 |
+
std_args.pretrained_model_name_or_path,
|
125 |
# tokenizer=tokenizer,
|
126 |
+
controlnet=std_controlnet,
|
127 |
safety_checker=None,
|
128 |
dtype=jnp.float32, # jnp.bfloat16
|
129 |
+
revision=std_args.revision,
|
130 |
+
from_pt=std_args.from_pt,
|
131 |
)
|
132 |
+
enc_pipeline, enc_pipeline_params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
|
133 |
+
enc_args.pretrained_model_name_or_path,
|
134 |
+
# tokenizer=tokenizer,
|
135 |
+
controlnet=enc_controlnet,
|
136 |
+
safety_checker=None,
|
137 |
+
dtype=jnp.float32, # jnp.bfloat16
|
138 |
+
revision=enc_args.revision,
|
139 |
+
from_pt=enc_args.from_pt,
|
140 |
+
)
|
141 |
+
|
142 |
|
143 |
+
std_pipeline_params["controlnet"] = std_controlnet_params
|
144 |
+
std_pipeline_params = jax_utils.replicate(std_pipeline_params)
|
145 |
|
146 |
+
enc_pipeline_params["controlnet"] = enc_controlnet_params
|
147 |
+
enc_pipeline_params = jax_utils.replicate(enc_pipeline_params)
|
148 |
|
149 |
rng = jax.random.PRNGKey(0)
|
150 |
num_samples = jax.device_count()
|
151 |
prng_seed = jax.random.split(rng, jax.device_count())
|
152 |
|
153 |
|
154 |
+
def infer(prompt, negative_prompt, image, model_type="Standard"):
|
155 |
prompts = num_samples * [prompt]
|
156 |
+
if model_type=="Standard":
|
157 |
+
prompt_ids = std_pipeline.prepare_text_inputs(prompts)
|
158 |
+
if model_type=="Hand Encoding":
|
159 |
+
prompt_ids = enc_pipeline.prepare_text_inputs(prompts)
|
160 |
prompt_ids = shard(prompt_ids)
|
161 |
|
162 |
if model_type=="Standard":
|
|
|
166 |
annotated_image = generate_annotation(image, overlap=False, hand_encoding=True)
|
167 |
overlap_image = generate_annotation(image, overlap=True, hand_encoding=True)
|
168 |
validation_image = Image.fromarray(annotated_image).convert("RGB")
|
|
|
|
|
169 |
|
170 |
+
if model_type=="Standard":
|
171 |
+
processed_image = std_pipeline.prepare_image_inputs(num_samples * [validation_image])
|
172 |
+
processed_image = shard(processed_image)
|
173 |
+
|
174 |
+
negative_prompt_ids = std_pipeline.prepare_text_inputs([negative_prompt] * num_samples)
|
175 |
+
negative_prompt_ids = shard(negative_prompt_ids)
|
176 |
+
|
177 |
+
images = std_pipeline(
|
178 |
+
prompt_ids=prompt_ids,
|
179 |
+
image=processed_image,
|
180 |
+
params=std_pipeline_params,
|
181 |
+
prng_seed=prng_seed,
|
182 |
+
num_inference_steps=50,
|
183 |
+
neg_prompt_ids=negative_prompt_ids,
|
184 |
+
jit=True,
|
185 |
+
).images
|
186 |
+
if model_type=="Hand Encoding":
|
187 |
+
processed_image = enc_pipeline.prepare_image_inputs(num_samples * [validation_image])
|
188 |
+
processed_image = shard(processed_image)
|
189 |
|
190 |
+
negative_prompt_ids = enc_pipeline.prepare_text_inputs([negative_prompt] * num_samples)
|
191 |
+
negative_prompt_ids = shard(negative_prompt_ids)
|
192 |
+
|
193 |
+
images = enc_pipeline(
|
194 |
+
prompt_ids=prompt_ids,
|
195 |
+
image=processed_image,
|
196 |
+
params=enc_pipeline_params,
|
197 |
+
prng_seed=prng_seed,
|
198 |
+
num_inference_steps=50,
|
199 |
+
neg_prompt_ids=negative_prompt_ids,
|
200 |
+
jit=True,
|
201 |
+
).images
|
202 |
|
203 |
|
204 |
images = images.reshape((images.shape[0] * images.shape[1],) + images.shape[-3:])
|
|
|
215 |
Model1 can be found at [https://huggingface.co/Vincent-luo/controlnet-hands](https://huggingface.co/Vincent-luo/controlnet-hands)
|
216 |
|
217 |
Model2 can be found at [https://huggingface.co/MakiPan/controlnet-encoded-hands-130k/ ](https://huggingface.co/MakiPan/controlnet-encoded-hands-130k/)
|
|
|
218 |
Dataset1 can be found at [https://huggingface.co/datasets/MakiPan/hagrid250k-blip2](https://huggingface.co/datasets/MakiPan/hagrid250k-blip2)
|
219 |
|
220 |
Dataset2 can be found at [https://huggingface.co/datasets/MakiPan/hagrid-hand-enc-250k](https://huggingface.co/datasets/MakiPan/hagrid-hand-enc-250k)
|
221 |
|
222 |
Preprocessing1 can be found at [https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/normal-preprocessing.py](https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/normal-preprocessing.py)
|
|
|
223 |
Preprocessing2 can be found at [https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/Hand-encoded-preprocessing.py](https://github.com/Maki-DS/Jax-Controlnet-hand-training/blob/main/Hand-encoded-preprocessing.py)
|
224 |
""")
|
225 |
+
model_type = gr.Radio(["Standard", "Hand Encoding"], label="Model preprocessing", info="We developed two models, one with standard mediapipe landmarks, and one with different (but similar) coloring on palm landmards to distinguish left and right")
|
226 |
+
|
227 |
with gr.Row():
|
228 |
with gr.Column():
|
229 |
prompt_input = gr.Textbox(label="Prompt")
|
|
|
265 |
"example4.png"
|
266 |
],
|
267 |
],
|
268 |
+
inputs=[prompt_input, negative_prompt, input_image, model_type],
|
269 |
outputs=[output_image],
|
270 |
fn=infer,
|
271 |
cache_examples=True,
|
272 |
)
|
273 |
|
274 |
+
inputs = [prompt_input, negative_prompt, input_image, model_type]
|
275 |
submit_btn.click(fn=infer, inputs=inputs, outputs=[output_image])
|
276 |
|
277 |
demo.launch()
|