Spaces:
Running
on
Zero
Running
on
Zero
parokshsaxena
commited on
Commit
β’
8c02e1d
1
Parent(s):
b608c7b
handling both shein images and viton standard images
Browse files
app.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
import logging
|
|
|
2 |
import gradio as gr
|
3 |
from PIL import Image
|
4 |
from src.tryon_pipeline import StableDiffusionXLInpaintPipeline as TryonPipeline
|
@@ -135,16 +136,25 @@ CATEGORY = "upper_body" # "lower_body"
|
|
135 |
|
136 |
@spaces.GPU
|
137 |
def start_tryon(dict,garm_img,garment_des, background_img, is_checked,is_checked_crop,denoise_steps,seed):
|
138 |
-
device = "cuda"
|
139 |
-
|
140 |
|
141 |
openpose_model.preprocessor.body_estimation.model.to(device)
|
142 |
pipe.to(device)
|
143 |
pipe.unet_encoder.to(device)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
144 |
|
145 |
garm_img= garm_img.convert("RGB").resize((WIDTH,HEIGHT))
|
146 |
-
# human_img_orig = dict["background"].convert("RGB")
|
147 |
-
human_img_orig = dict.convert("RGB")
|
148 |
if is_checked_crop:
|
149 |
width, height = human_img_orig.size
|
150 |
target_width = int(min(width, height * (3 / 4)))
|
@@ -272,8 +282,8 @@ human_list = os.listdir(os.path.join(example_path,"human"))
|
|
272 |
human_list_path = [os.path.join(example_path,"human",human) for human in human_list]
|
273 |
|
274 |
human_ex_list = []
|
275 |
-
human_ex_list = human_list_path
|
276 |
-
""" if using ImageEditor instead of Image while taking input, use this
|
277 |
for ex_human in human_list_path:
|
278 |
ex_dict= {}
|
279 |
ex_dict['background'] = ex_human
|
@@ -293,7 +303,7 @@ with image_blocks as demo:
|
|
293 |
with gr.Column():
|
294 |
# changing from ImageEditor to Image to allow easy passing of data through API
|
295 |
# instead of passing {"dictionary": <>} ( which is failing ), we can directly pass the image
|
296 |
-
#
|
297 |
imgs = gr.Image(sources='upload', type='pil',label='Human. Mask with pen or use auto-masking')
|
298 |
with gr.Row():
|
299 |
is_checked = gr.Checkbox(label="Yes", info="Use auto-generated mask (Takes 5 seconds)",value=True)
|
|
|
1 |
import logging
|
2 |
+
import math
|
3 |
import gradio as gr
|
4 |
from PIL import Image
|
5 |
from src.tryon_pipeline import StableDiffusionXLInpaintPipeline as TryonPipeline
|
|
|
136 |
|
137 |
@spaces.GPU
|
138 |
def start_tryon(dict,garm_img,garment_des, background_img, is_checked,is_checked_crop,denoise_steps,seed):
|
139 |
+
#device = "cuda"
|
140 |
+
device = 'cuda:0' if torch.cuda.is_available() else 'cpu'
|
141 |
|
142 |
openpose_model.preprocessor.body_estimation.model.to(device)
|
143 |
pipe.to(device)
|
144 |
pipe.unet_encoder.to(device)
|
145 |
+
|
146 |
+
#human_img_orig = dict["background"].convert("RGB") # ImageEditor
|
147 |
+
human_img_orig = dict.convert("RGB") # Image
|
148 |
+
|
149 |
+
# Derive HEIGHT & WIDTH such that width is not more than 1000. This will cater to both Shein images (4160x6240) of 3:4 AR and model standard images ( 768x1024 ) of 2:3 AR
|
150 |
+
WIDTH, HEIGHT = human_img_orig.size
|
151 |
+
division_factor = math.ceil(WIDTH/1000)
|
152 |
+
WIDTH = int(WIDTH/division_factor)
|
153 |
+
HEIGHT = int(HEIGHT/division_factor)
|
154 |
+
POSE_WIDTH = int(WIDTH/2)
|
155 |
+
POSE_HEIGHT = int(HEIGHT/2)
|
156 |
|
157 |
garm_img= garm_img.convert("RGB").resize((WIDTH,HEIGHT))
|
|
|
|
|
158 |
if is_checked_crop:
|
159 |
width, height = human_img_orig.size
|
160 |
target_width = int(min(width, height * (3 / 4)))
|
|
|
282 |
human_list_path = [os.path.join(example_path,"human",human) for human in human_list]
|
283 |
|
284 |
human_ex_list = []
|
285 |
+
human_ex_list = human_list_path # Image
|
286 |
+
""" if using ImageEditor instead of Image while taking input, use this - ImageEditor
|
287 |
for ex_human in human_list_path:
|
288 |
ex_dict= {}
|
289 |
ex_dict['background'] = ex_human
|
|
|
303 |
with gr.Column():
|
304 |
# changing from ImageEditor to Image to allow easy passing of data through API
|
305 |
# instead of passing {"dictionary": <>} ( which is failing ), we can directly pass the image
|
306 |
+
#imgs = gr.ImageEditor(sources='upload', type="pil", label='Human. Mask with pen or use auto-masking', interactive=True)
|
307 |
imgs = gr.Image(sources='upload', type='pil',label='Human. Mask with pen or use auto-masking')
|
308 |
with gr.Row():
|
309 |
is_checked = gr.Checkbox(label="Yes", info="Use auto-generated mask (Takes 5 seconds)",value=True)
|