Update app.py
Browse files
app.py
CHANGED
@@ -33,6 +33,8 @@ footer {
|
|
33 |
}
|
34 |
'''
|
35 |
|
|
|
|
|
36 |
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
|
37 |
controlnet = ControlNetModel.from_pretrained("MakiPan/controlnet-encoded-hands-130k", torch_dtype=torch.float16)
|
38 |
|
@@ -125,11 +127,15 @@ def generate(input=DEFAULT_INPUT, filter_input="", negative_input=DEFAULT_NEGATI
|
|
125 |
|
126 |
print(image_paths)
|
127 |
|
128 |
-
|
|
|
|
|
|
|
|
|
129 |
|
130 |
-
print(
|
131 |
|
132 |
-
return image_paths,
|
133 |
|
134 |
def cloud():
|
135 |
print("[CLOUD] | Space maintained.")
|
@@ -156,9 +162,9 @@ with gr.Blocks(css=css) as main:
|
|
156 |
|
157 |
with gr.Column():
|
158 |
images = gr.Gallery(columns=1, label="Image")
|
159 |
-
|
160 |
|
161 |
-
submit.click(generate, inputs=[input, filter_input, negative_input, model, height, width, steps, guidance, number, seed], outputs=[images,
|
162 |
maintain.click(cloud, inputs=[], outputs=[], queue=False)
|
163 |
|
164 |
main.launch(show_api=True)
|
|
|
33 |
}
|
34 |
'''
|
35 |
|
36 |
+
repo_nsfw_classifier = pipeline("image-classification", model="Falconsai/nsfw_image_detection")
|
37 |
+
|
38 |
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
|
39 |
controlnet = ControlNetModel.from_pretrained("MakiPan/controlnet-encoded-hands-130k", torch_dtype=torch.float16)
|
40 |
|
|
|
127 |
|
128 |
print(image_paths)
|
129 |
|
130 |
+
nsfw_prediction = repo_nsfw_classifier(Image.open(image_paths[0]))
|
131 |
+
|
132 |
+
print(nsfw_prediction)
|
133 |
+
|
134 |
+
nsfw_format = [f"{index['label']}: {index['score']:.3%}" for index in nsfw_prediction]
|
135 |
|
136 |
+
print(nsfw_format)
|
137 |
|
138 |
+
return image_paths, nsfw_format
|
139 |
|
140 |
def cloud():
|
141 |
print("[CLOUD] | Space maintained.")
|
|
|
162 |
|
163 |
with gr.Column():
|
164 |
images = gr.Gallery(columns=1, label="Image")
|
165 |
+
nsfw_classifier = gr.Label()
|
166 |
|
167 |
+
submit.click(generate, inputs=[input, filter_input, negative_input, model, height, width, steps, guidance, number, seed], outputs=[images, nsfw_classifier], queue=False)
|
168 |
maintain.click(cloud, inputs=[], outputs=[], queue=False)
|
169 |
|
170 |
main.launch(show_api=True)
|