Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,18 +1,14 @@
|
|
1 |
from typing import List
|
2 |
-
|
3 |
import gradio as gr
|
4 |
import numpy as np
|
5 |
import supervision as sv
|
6 |
import torch
|
7 |
-
|
8 |
from PIL import Image
|
9 |
-
from transformers import pipeline
|
10 |
-
|
11 |
|
12 |
-
|
13 |
-
#Variables globales
|
14 |
MARKDOWN = """
|
15 |
-
#SAM
|
16 |
"""
|
17 |
EXAMPLES = [
|
18 |
["https://media.roboflow.com/notebooks/examples/dog.jpeg", "dog", 0.5],
|
@@ -22,130 +18,79 @@ EXAMPLES = [
|
|
22 |
]
|
23 |
|
24 |
MIN_AREA_THRESHOLD = 0.01
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
-
|
27 |
-
|
28 |
-
SAM_GENERATOR = pipeline(
|
29 |
-
task = "mask-generation",
|
30 |
-
model = "facebook/sam-vit-large",
|
31 |
-
device = DEVICE
|
32 |
-
)
|
33 |
-
|
34 |
SEMITRANSPARENT_MASK_ANNOTATOR = sv.MaskAnnotator(
|
35 |
-
color
|
36 |
-
color_lookup
|
37 |
)
|
38 |
|
39 |
SOLID_MASK_ANNOTATOR = sv.MaskAnnotator(
|
40 |
-
color
|
41 |
-
color_lookup
|
42 |
-
opacity
|
43 |
)
|
44 |
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
|
55 |
def reverse_mask_image(image: np.ndarray, mask: np.ndarray, gray_value=128):
|
56 |
-
gray_color = np.array([
|
57 |
-
gray_value,
|
58 |
-
gray_value,
|
59 |
-
gray_value
|
60 |
-
], dtype=np.uint8)
|
61 |
return np.where(mask[..., None], image, gray_color)
|
62 |
|
63 |
-
|
64 |
-
"""
|
65 |
-
def filter_detections(image_rgb_pil: Image.Image, detections: sv.Detections) -> sv.Detections:
|
66 |
-
img_rgb_numpy = np.array(image_rgb_pil)
|
67 |
-
filtering_mask = []
|
68 |
-
for xyxy, mask in zip(detections.xyxy, detections.mask):
|
69 |
-
crop = sv.crop_image(
|
70 |
-
image = img_rgb_numpy,
|
71 |
-
xyxy =xyxy
|
72 |
-
)
|
73 |
-
mask_crop = sv.crop_image(
|
74 |
-
image=mask,
|
75 |
-
xyxy=xyxy
|
76 |
-
)
|
77 |
-
masked_crop = reverse_mask_image(
|
78 |
-
image=crop,
|
79 |
-
mask=mask_crop
|
80 |
-
)
|
81 |
-
|
82 |
-
filtering_mask = np.array(
|
83 |
-
filtering_mask
|
84 |
-
)
|
85 |
-
return detections[filtering_mask]
|
86 |
-
"""
|
87 |
-
|
88 |
-
def inference (image_rgb_pil: Image.Image) -> List[Image.Image]:
|
89 |
width, height = image_rgb_pil.size
|
90 |
area = width * height
|
91 |
|
92 |
-
detections = run_sam(
|
93 |
-
|
94 |
-
|
95 |
-
detections = detections[ detections.area /area > MIN_AREA_THRESHOLD ]
|
96 |
-
|
97 |
-
#detections = filter_detections(
|
98 |
-
# image_rgb_pil=image_rgb_pil,
|
99 |
-
# detections=detections,
|
100 |
-
#)
|
101 |
-
|
102 |
blank_image = Image.new("RGB", (width, height), "black")
|
103 |
return [
|
104 |
-
annotate(
|
105 |
-
|
106 |
-
detections=detections,
|
107 |
-
annotator=SEMITRANSPARENT_MASK_ANNOTATOR),
|
108 |
-
annotate(
|
109 |
-
image_rgb_pil=blank_image,
|
110 |
-
detections=detections,
|
111 |
-
annotator=SOLID_MASK_ANNOTATOR)
|
112 |
]
|
113 |
-
|
114 |
-
|
115 |
#************
|
116 |
#GRADIO CONSTRUCTION
|
117 |
with gr.Blocks() as demo:
|
118 |
gr.Markdown(MARKDOWN)
|
119 |
with gr.Row():
|
120 |
with gr.Column():
|
121 |
-
input_image = gr.Image(
|
122 |
-
image_mode = 'RGB',
|
123 |
-
type = 'pil',
|
124 |
-
height = 500
|
125 |
-
)
|
126 |
submit_button = gr.Button("Pruébalo!!!")
|
127 |
-
gallery = gr.Gallery(
|
128 |
-
|
129 |
-
object_fit = "scale-down",
|
130 |
-
preview = True
|
131 |
-
)
|
132 |
with gr.Row():
|
133 |
gr.Examples(
|
134 |
-
examples
|
135 |
-
fn
|
136 |
-
inputs
|
137 |
-
|
138 |
-
|
139 |
-
|
140 |
-
cache_examples = False,
|
141 |
-
run_on_click = True
|
142 |
)
|
143 |
submit_button.click(
|
144 |
inference,
|
145 |
-
inputs
|
146 |
-
|
147 |
-
],
|
148 |
-
outputs = gallery
|
149 |
)
|
150 |
|
151 |
-
demo.launch(
|
|
|
1 |
from typing import List
|
|
|
2 |
import gradio as gr
|
3 |
import numpy as np
|
4 |
import supervision as sv
|
5 |
import torch
|
|
|
6 |
from PIL import Image
|
7 |
+
from transformers import pipeline
|
|
|
8 |
|
9 |
+
# Global Variables
|
|
|
10 |
MARKDOWN = """
|
11 |
+
# SAM - Softly Activated Masks
|
12 |
"""
|
13 |
EXAMPLES = [
|
14 |
["https://media.roboflow.com/notebooks/examples/dog.jpeg", "dog", 0.5],
|
|
|
18 |
]
|
19 |
|
20 |
MIN_AREA_THRESHOLD = 0.01
|
21 |
+
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
|
22 |
+
|
23 |
+
# Initialize SAM Generator with exception handling
|
24 |
+
try:
|
25 |
+
SAM_GENERATOR = pipeline(
|
26 |
+
task="mask-generation",
|
27 |
+
model="facebook/sam-vit-large",
|
28 |
+
device=DEVICE
|
29 |
+
)
|
30 |
+
except Exception as e:
|
31 |
+
print(f"Error initializing SAM generator: {e}")
|
32 |
|
33 |
+
# Mask Annotators
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
SEMITRANSPARENT_MASK_ANNOTATOR = sv.MaskAnnotator(
|
35 |
+
color=sv.Color.red(),
|
36 |
+
color_lookup=sv.ColorLookup.INDEX
|
37 |
)
|
38 |
|
39 |
SOLID_MASK_ANNOTATOR = sv.MaskAnnotator(
|
40 |
+
color=sv.Color.white(),
|
41 |
+
color_lookup=sv.ColorLookup.INDEX,
|
42 |
+
opacity=1
|
43 |
)
|
44 |
|
45 |
+
# Functions
|
46 |
+
def run_sam(image_rgb_pil: Image.Image) -> sv.Detections:
|
47 |
+
try:
|
48 |
+
outputs = SAM_GENERATOR(image_rgb_pil, points_per_batch=32)
|
49 |
+
mask = np.array(outputs['masks'], dtype=np.uint8)
|
50 |
+
return sv.Detections(xyxy=sv.mask_to_xyxy(masks=mask), mask=mask)
|
51 |
+
except Exception as e:
|
52 |
+
print(f"Error running SAM model: {e}")
|
53 |
+
return sv.Detections(xyxy=[], mask=[])
|
54 |
|
55 |
def reverse_mask_image(image: np.ndarray, mask: np.ndarray, gray_value=128):
|
56 |
+
gray_color = np.array([gray_value, gray_value, gray_value], dtype=np.uint8)
|
|
|
|
|
|
|
|
|
57 |
return np.where(mask[..., None], image, gray_color)
|
58 |
|
59 |
+
def inference(image_rgb_pil: Image.Image) -> List[Image.Image]:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
width, height = image_rgb_pil.size
|
61 |
area = width * height
|
62 |
|
63 |
+
detections = run_sam(image_rgb_pil)
|
64 |
+
detections = detections[detections.area / area > MIN_AREA_THRESHOLD]
|
65 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
66 |
blank_image = Image.new("RGB", (width, height), "black")
|
67 |
return [
|
68 |
+
SEMITRANSPARENT_MASK_ANNOTATOR.annotate(image_rgb_pil, detections),
|
69 |
+
SOLID_MASK_ANNOTATOR.annotate(blank_image, detections)
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
]
|
|
|
|
|
71 |
#************
|
72 |
#GRADIO CONSTRUCTION
|
73 |
with gr.Blocks() as demo:
|
74 |
gr.Markdown(MARKDOWN)
|
75 |
with gr.Row():
|
76 |
with gr.Column():
|
77 |
+
input_image = gr.Image(image_mode='RGB', type='pil', height=500)
|
|
|
|
|
|
|
|
|
78 |
submit_button = gr.Button("Pruébalo!!!")
|
79 |
+
gallery = gr.Gallery(label="Result", object_fit="scale-down", preview=True)
|
80 |
+
|
|
|
|
|
|
|
81 |
with gr.Row():
|
82 |
gr.Examples(
|
83 |
+
examples=EXAMPLES,
|
84 |
+
fn=inference,
|
85 |
+
inputs=[input_image],
|
86 |
+
outputs=[gallery],
|
87 |
+
cache_examples=False,
|
88 |
+
run_on_click=True
|
|
|
|
|
89 |
)
|
90 |
submit_button.click(
|
91 |
inference,
|
92 |
+
inputs=[input_image],
|
93 |
+
outputs=gallery
|
|
|
|
|
94 |
)
|
95 |
|
96 |
+
demo.launch(debug=False, show_error=True)
|