Spaces:
Sleeping
Sleeping
Kunpeng Song
commited on
Commit
·
a6600be
1
Parent(s):
5b2734d
fix zero
Browse files- .DS_Store +0 -0
- app.py +6 -2
- dataset_lib/dataset_eval_MoMA.py +2 -2
.DS_Store
CHANGED
Binary files a/.DS_Store and b/.DS_Store differ
|
|
app.py
CHANGED
@@ -1,12 +1,13 @@
|
|
1 |
import spaces
|
2 |
-
|
3 |
import gradio as gr
|
4 |
import torch
|
5 |
import numpy as np
|
6 |
import torch
|
7 |
from pytorch_lightning import seed_everything
|
8 |
from model_lib.utils import parse_args
|
9 |
-
import
|
|
|
10 |
os.environ["CUDA_VISIBLE_DEVICES"]="0"
|
11 |
|
12 |
title = "MoMA"
|
@@ -17,6 +18,9 @@ args = parse_args()
|
|
17 |
|
18 |
model = None
|
19 |
|
|
|
|
|
|
|
20 |
@spaces.GPU
|
21 |
def inference(rgb, subject, prompt, strength, seed):
|
22 |
seed = int(seed) if seed else 0
|
|
|
1 |
import spaces
|
2 |
+
import os
|
3 |
import gradio as gr
|
4 |
import torch
|
5 |
import numpy as np
|
6 |
import torch
|
7 |
from pytorch_lightning import seed_everything
|
8 |
from model_lib.utils import parse_args
|
9 |
+
from llava.mm_utils import process_image
|
10 |
+
|
11 |
os.environ["CUDA_VISIBLE_DEVICES"]="0"
|
12 |
|
13 |
title = "MoMA"
|
|
|
18 |
|
19 |
model = None
|
20 |
|
21 |
+
def my_process_image(a, b, c):
|
22 |
+
return process_image(a, b, c)
|
23 |
+
|
24 |
@spaces.GPU
|
25 |
def inference(rgb, subject, prompt, strength, seed):
|
26 |
seed = int(seed) if seed else 0
|
dataset_lib/dataset_eval_MoMA.py
CHANGED
@@ -2,7 +2,7 @@ from PIL import Image
|
|
2 |
import numpy as np
|
3 |
import torch
|
4 |
from torchvision import transforms
|
5 |
-
from
|
6 |
from rembg import remove
|
7 |
|
8 |
def create_binary_mask(image):
|
@@ -38,7 +38,7 @@ def Dataset_evaluate_MoMA(image_pil, prompt,subject, moMA_main_modal):
|
|
38 |
image_wb = image * mask + torch.ones_like(image)* (1-mask)*255
|
39 |
image_pil = Image.fromarray(image_wb.permute(1,2,0).numpy().astype(np.uint8))
|
40 |
|
41 |
-
res['llava_processed'] =
|
42 |
res['label'] = [subject]
|
43 |
return res
|
44 |
|
|
|
2 |
import numpy as np
|
3 |
import torch
|
4 |
from torchvision import transforms
|
5 |
+
from ..app import my_process_image
|
6 |
from rembg import remove
|
7 |
|
8 |
def create_binary_mask(image):
|
|
|
38 |
image_wb = image * mask + torch.ones_like(image)* (1-mask)*255
|
39 |
image_pil = Image.fromarray(image_wb.permute(1,2,0).numpy().astype(np.uint8))
|
40 |
|
41 |
+
res['llava_processed'] = my_process_image([image_pil], LLaVa_processor, llava_config)
|
42 |
res['label'] = [subject]
|
43 |
return res
|
44 |
|